Introduction
Statistical analysis proves particularly useful when making organizational decisions. Decisions based on biases or personal opinions may show only isolated effectiveness, whereas data for making recommendations is an objective and unbiased source (Simeunović et al., 2019). Consequently, when a business turns to a statistical test, the reliability of the results increases, and if implemented successfully, the performance and well-being of the corporate environment can improve. This paper proposes using the FRINGE dataset to determine the relationships and patterns that are fair to the information collected.
Research Questions and Hypotheses
To conduct the statistical analysis, five research questions were initially formulated to create the conceptual framework for this project. Specifically, the following questions were relevant:
- What impact does gender have on employees’ annual earnings?
- Is work experience with a given employer related to the employee’s annual earnings level?
- How does the type of industry affect the value of employee insurance?
- Does the number of years worked in interaction with age affect the value of vacation days?
- How accurately do hourly earnings and the number of dependents predict office work?
The questions formulated will allow us to examine the patterns in the data and determine the relationships among the variables. Because any inferential statistics test must be based on hypothesis testing, it was necessary to identify null and alternative hypotheses for each research question. For convenience, all hypotheses are summarized in the table below.
Methods
IBM SPSS was used to perform statistical analyses, allowing for automatic calculations. The original variables were initially recoded for the tests (KSU, 2023). In particular, the categorical variables were initially generated as separate variables, so recoding them into aggregates was necessary. An independent samples t-test was used to test the first question, and Pearson’s correlation was used for the second question.
The third research question was tested through a one-way ANOVA, as there were nine levels of the independent variable and one continuous dependent variable. Multiple regression was used to test the fourth question, taking into account the interaction factor between the variables (Hu & Plonsky, 2021). Finally, multiple regression with a probit model and interaction factor was used to test the fifth question.
Dataset
The proposed FRINGE dataset, with a total of 616 records, consisted of 39 variables, which were reduced to 29 after the recoding procedure. Given the stated research questions, after removing unused variables, the number of variables was 11, three of which (gender, office work, and industry type) were measured on a nominal scale. This section offers a theoretical framework for formulating each hypothesis.
Gender and Earnings
The study’s hypotheses assume that gender has a significant impact on earnings levels. This assumption for the sample is based on the phenomenon of the gender wage gap, according to which men tend to earn more money than women (Petrongolo, 2019). According to Aragão (2023), for every $1 earned by a man, a woman earned only 82 cents. This creates a policy of gender inequality and discrimination in which women and men with equal skills earn different salaries. Thus, this hypothesis aimed to investigate the gender wage gap phenomenon using the sample data.
Work Experience and Earnings
The hypothesis suggests a significant correlation between work experience and an employee’s annual earnings. The hypothesis is based on the idea that the more a person works for a company, the more in-depth and relevant knowledge they have, which leads to higher wages (IET, 2022). Priya et al. (2020) demonstrated that increased work experience leads to higher wages for employees. Thus, work experience can increase wages, and this assumption was interesting to test for the sample data.
Type of Industry and Value of Insurance
The hypothesis suggests that the value of insurance varies significantly across different industries. The hypothesis assumes that employers make insurance payments, and companies in different industries may create different insurance terms for employees (SHRM, 2021; Smith, 2022). In testing this hypothesis, it was interesting to investigate whether there was a single significant difference in the sample averages or whether all industries offered the same insurance terms to employees
Number of Years Earned, Age, and Value of Vacation Days
The hypothesis for this study posited that leave was predictable based on the number of years an employee had worked, their age, and the interaction between these two factors. Generally, while a higher value of vacation days increases the tangible benefit for the employee, it simultaneously requires the employer to allocate a greater amount of money.
Given that higher work experience is associated with higher wages, as Priya et al. (2020) suggested, this also leads to a higher value of the employee’s vacation days. Age may also have influenced this indicator, as older employees tend to have more work experience due to age. The interaction factor of the two independent variables would determine whether there is a significant effect of the number of years worked and age on the vacation characteristic.
Hourly Earnings, Number of Dependents, and Office Work
This hypothesis hypothesized that hourly earnings, the number of dependents, and their interaction might influence the decision to work in an office. This hypothesis is based on the phenomenon of remote working, which has intensified markedly since the COVID-19 pandemic (Espitia et al., 2022). Employees can often decide whether to work in the office or perform the same tasks from home.
Hourly earnings were used as one of the independent variables because it was hypothesized that this indicator may have influenced individuals’ choice of work format (Davies, 2020). At the same time, the number of dependents may also have had an influence, as the more dependents an individual has, the more likely it seems that a remote work format is appropriate for them (Kelly et al., 2023). Finally, the interaction factor of these two variables was used to assess the combined effect of hourly earnings and the number of dependents at different levels.
Data Analysis and Discussion
Descriptive Statistics
The sample consisted of 616 respondents, of whom 35.9% (n = 221) were female, and 64.1% (n = 395) were male. In terms of age distribution, the mean age of the sample was 37.65 years (SD = 12.70), with the oldest respondent being 80 years old and the youngest respondent being 16 years old. Average hourly earnings were $6.23 (SD = $4.84), and average annual earnings were $12,887.72 (SD = $9,013.01). Respondents had an average of 18.62 years of work experience (SD = 12.33), with a non-changing employer, which was expectedly lower (M = 7.75, SD = 7.78).
The average value of an individual’s insurance was $569.00 (SD = $421.42). The sample represented 51.6% (n = 318) of individuals who worked out of the office and 48.4% (n = 298) of those who worked in the office. Regarding industrial distribution, Figure 1 shows that Industries 3 and 9 had the highest number of employees (20.62%, n = 127), and Industry 1 had the lowest number (0.32%, n = 2) of employees in the sample. Finally, in terms of having dependents, respondents had an average of 1.23 (SD = 1.41) dependents, with a maximum of 7.
Gender and Earnings
A t-test of independent samples showed a significant difference in earnings between women and men (t(592.08) = -10.594, p <.001). The result indicates that men (M = $15,306.88, SD = $9,376.66) actually earned more per year than women (M = $8,563.88, SD = $6,351.28). This finding aligns well with the well-known phenomenon of the gender wage gap, indicating that for the sample, the presence of such a gap is not exceptional (Aragão, 2023). However, while Aragão reported that women earn up to 82% of men’s wages, the sample results show a higher gap, specifically, for every dollar earned in a year by a man, the woman in the sample earned only 56 cents.
Work Experience and Earnings Level
The correlation results showed a significant relationship (r =.129, p =.001) between work experience with a given employer and wage level. Despite the significance of the correlation, the relationship detected was rather weak and not pronounced. At this point, an interesting addition was the assessment of the correlation between total work experience and annual earnings: the data showed a stronger correlation (r =.281, p <.001). This implied that the correlation was significant in both cases, but the relationship was more pronounced in the case of general experience than in that of experience with a particular employer.
Overall, the data supported the findings of previously published studies. Both Priya et al. (2020) and IET (2022) demonstrated an association between increased experience and salary growth, with similar results observed in the sample data. However, the weak relationship should be considered, indicating the potential for a more complex relationship and linkage to additional variables not included in the current correlation analysis.
Industry Type and Value of Insurance
The results showed significant differences (F(8, 607) = 35.618, p <.001), and an additional Tukey post hoc test was conducted to analyze the locations of these differences. This determined that differences existed among virtually all industries. Thus, the amount of coverage for Industry 2 (M = $675.67, SD = $447.14) was significantly lower than that for Industry 3 (M = $927.11, SD = $481. 82) but higher than for Industry 6 (M = $359.57, SD = $307.89), Industry 7 (M = $423.79, SD = $192.13), Industry 8 (M = $326.92, SD = $284.80), and Industry 9 (M = $333.03, SD = $231.03).
However, no significant differences were found between Industries 6, 7, 8, and 9. Industry 3 had significantly higher employee coverage than Industry 4 (M = $665.27, SD = $341.72) and Industries 6, 7, 8, and 9. Differences were also found for Industry 5 (M = $782.57, SD = $351.08), with higher rates than the last four industries. Thus, Industries 3 and 5 had the maximum insurance coverage for employees, while Industries 6, 7, 8, and 9 had the lowest rates.
From the results, it is evident that the value of insurance varied between industries, which supports SHRM’s (2021) and Smith’s (2022) findings that employers are entitled to offer unequal employee insurance packages. Higher amounts of insurance appear to be more attractive to employees because they can qualify for larger benefits; however, it also means that the employer withholds a larger portion of wages to direct them to insurance funds.
Number of Years Earned, Age, and Value of Vacation Days
The results showed a significant linear model (F(3, 612) = 29.753, p <.001) with a coefficient of determination of.123. This indicates low model reliability but significant results, so they should be interpreted cautiously. Work experience (β = 1.460, p <.001) and interaction (β = -1.343, p <.001) coefficients had a significant effect, whereas the age of individuals (β = 0.069, p =.466) did not. Without the inclusion of the interaction factor, however, the coefficients of work experience (β = 12.469, p <.001) and age (β = -6.896, p =.005) are significant, but the explanatory potential of the model is reduced to R2 =.054.
It follows that an increase in work experience has a positive effect on the value of vacation days, which is consistent with the assumptions previously described. However, the individual’s age had no effect on this indicator when the interaction term was included, which means that older employees cannot be said to have a higher vacation value in this model. However, without considering the interaction factor, age was a significant predictor of the value of vacation days, resulting in its reduction. The results found for the interaction factor indicate that, depending on age, the value of vacation days decreases as experience increases.
Hourly Earnings, Number of Dependents, and Working in the Office
Multiple regression with an interaction factor was used to analyze the relationships between the above variables, with the difference being that the dependent variable was measured on a binary scale. Thus, the regression model had to be built on a probit algorithm. The results showed a significant regression model (χ2(3) = 24.750, p <.001), which means that the model demonstrates the significance of predictor effects on office job choice.
The number of dependents was a significant factor (B = -0.183, p =.020), indicating that the more dependents an employee has, the more likely they are to refuse the office job option. The coefficient on hourly earnings was also significant (B = 0.043, p =.018), indicating that the higher an employee’s earnings, the more likely they were to work in an office setting. Meanwhile, the interaction coefficient (B = 0.018, p =.116) was not a significant predictor, indicating that these variables were independent in influencing office work.
The results support the assumption that employees with more dependents are less likely to choose to work in the office, as shown in Kelly et al. (2023). It was also found that employees with higher hourly wages were more likely to work in an office: this is coupled with the fact that office workers tend to be paid more than remote workers (Davies, 2020). The lack of significance of the interaction factor demonstrated that there is no pattern between the two variables, and thus, they should be viewed as independent of each other.
Conclusion
Using statistical analysis, this paper focused on investigating patterns and relationships in a proposed dataset related to employee benefits. Five research questions and associated hypotheses were formulated, statistical tests were run, and results were obtained. A gender wage gap existed, with men earning more than women. The relationship between work experience and annual earnings was confirmed: employees with more work experience received more money.
Average insurance coverage differed significantly across almost all nine industries, with the highest values found in Industries 3 and 5. While working years and age initially had opposite effects on vacation day value, age became insignificant when the interaction factor was included. Similarly, hourly earnings and number of dependents showed opposite effects on the preference for remote versus office work, though their interaction was not a significant predictor. The overall findings offer valuable, practical insights for corporate well-being and organizational decisions.
References
Aragão, C. (2023, March 01). Gender pay gap in U.S. hasn’t changed much in two decades. Pew Research Center.
Davies, N. (2020). Should remote workers really be paid less than those in the office? Forbes.
Espitia, A., Mattoo, A., Rocha, N., Ruta, M., & Winkler, D. (2022). Pandemic trade: COVID‐19, remote work, and global value chains. The World Economy, 45(2), 561-589.
Hu, Y., & Plonsky, L. (2021). Statistical assumptions in L2 research: A systematic review. Second Language Research, 37(1), 171-184.
IET. (2022). What is a tenured employee? Benefits of earning this status. Indeed.
Kelly, J. A., Kelleher, L., Guo, Y., Deegan, C., Larsen, B., Shukla, S., & Collins, A. (2022). Assessing preference and potential for working from anywhere: A spatial index for Ireland. Environmental and Sustainability Indicators, 15, 1-23.
KSU. (2023). SPSS tutorials: Recoding variables. Kent State University.
Petrongolo, B. (2019). The gender gap in employment and wages. Nature Human Behaviour, 3(4), 316-318.
Priya, M. S. R., Lakshmi, P., & Dixit, S. (2020). Exploring the relationship between job tenure and salary: An empirical analysis. International Journal of Management (IJM), 11(7), 472-484.
SHRM. (2021). Are employers allowed to offer different benefits to different employees and to charge more for the same benefit, or is this a discriminatory practice? SHRM.
Simeunović, I., Vukajlović, V., Beraha, I., & Brzaković, M. (2019). Importance of information in crisis management – Statistical analysis. Industrija, 47(3), 1-17.
Smith, G. (2022). Can an employer contribute different amounts toward employee medical insurance? People Keep.