Impact of Part-time Jobs on Academic Performance

At present, the combination of academic study along with employment has become the norm for several students. The reasons for students to choose work along with study are many. Some have to bear the expenses of higher studies, while some desire to integrate into the job market while some others just do it for the sake of spending spare time. There are different ways in which the part time jobs of the students can influence the results of their education (Robotham, 2012). This can help in the development of specific personal characteristics, organisations and work, management of time and the enhancement of school achievements. It is however believed that the concept of employment reduces the available time for education and classes. management, in case students concentrate on their work, the time they get for studies is bound to get reduced and therefore might lead to low educational achievements, which might even cause them to receive notifications from their classes to leave (Koch, 2013). Motivation for working and perceptions of the jobs for the students develop the academic performance of the students (Source: “The effects of part-time work on school students”, 2018). Students as well as their parents and professors should be careful about the pros and cons of part time job after school-hours. The influencing factors regarding part time jobs should be found out and thoroughly analysed (Source: “Part-Time Work and Student Achievement – Educational Leadership”, 2018).

Therefore, the aim of this study is to understand the relationship between having a part-time job and academic performance of the students (Bryman,2015) . A quantitative research is carried out in this particular study and information is collected from 42 respondents who are basically students who are students of the University of Life Sciences. The method of collection of data is a questionnaire. The copy of the survey questionnaire which is used is to be attached with this particular report. The software which is being used for this particular project is R software.

Thesis Statement: The students who do part time jobs and their opinions on their impact on Academic performance

Research Question: How does having a part time job along with studies affect the academic performance of the students?

Objective: The aim of the study is to analyse about students who perform part time jobs and their impact on the performance of the students.

This section provides information on the theories which are utilized to achieve the targets of the study. The theories which are used in relation to the variables which are being considered with respect to the present study are to be considered in this particular case. A total of 42 students were provided with questionnaires in order to understand their responses relating to the impact of part time jobs on educational performance (Murphy, Myors &Wolach, 2014). The students from whom the data were collected studied in the University of Life Sciences. Students consider that they usually need to properly balance their jobs and their study time so that the part time jobs do not influence their time of study (Koch, 2013). They feel that in case the study time and job responsibilities need to be prioritised and segregated from each other at suitable times (Macan, 1990). Academic performance in this particular instance is the dependent variable which is impacted by the part times jobs in a positive or a negative way. The other variables which are to be considered for analysis in this particular case study are “the hours of lecture attended”, “gender”, “nationality”, “study level” and “work clash with class”.  It also needs to be analysed whether the introduction of the variables “gender”, “nationality”, “study level” shows significant changes in the results of the study than when only the two variables “part time job” and “the hours of lecture attended” are considered.

Theoretical Background

The two variables “par time job” and “the hours of lecture attended per week” are included in the three cases in order to understand the effect of the addition of the variables.

Null Hypothesis H0: There is no impact of having part time jobs on student performance in university

Alternative Hypothesis H1: There is a considerable and significant impact of having a part time job on the performance in the university.  

The method of quantitative research is utilized in this particular survey. The methods of analysis which are adopted are linear regression and multivariate regression. These methods are adopted in order to understand the impact of the part time jobs on the academic performance of the students. The statistical software R is utilized to obtain the results of this particular survey in order to understand the relationship between the variables. Here the dependent variable under consideration is academic performance. The independent variables are gender, nationality, study level, hours of class with the classes. (Bryman, 2017)

Data analysis is done in the R software to help understand the results. The OLS (ordinary least square technique) is utilized in statistical analysis as it helps in the interpretation of the regression coefficient and reduction of bias.

The source of the data is the students who study at University of life sciences in Australia. The researchers (three students as a group) have collected the data from the students who study at university of life sciences. The method of sampling for collecting the data set is “Survey questionnaire” method. This survey questionnaire has 16 questions about the topic of the research including demographic aspects of the responders. The focus of the survey is to find clear data about whether students do part time jobs or not and the academic performance of the students. The data collection process started in March, 2018 and took approximately ten days to collect the data. The researcher before collecting data, carried out a simple pilot test to test whether there are repetitions or not. The data collection technique assesses the validity of the questions relating to current research questions and hypotheses. The survey contains closed-ended questions. The questions were framed simple and clear for the participants. The responses after collecting are transformed in dummy variables. Researcher took the help of “Google forms” to design. The researcher provided a brief discussion of the study to clearly inform the respondents. Total 42 responders responded to this question. Some missing data is found due to null responses. The data is collected using email. The process is little bit of time-taking. According to the informed consent, we assumed to answer our survey consent.  

Methodology

The convenience-sampling technique is the sampling strategy of the report. The researcher utilised the convenience sampling because of accessibility to respondents. Therefore, the sampling strategy does not represent the entire population. The results of the study are not the representation of the whole study. The best sampling method which is simple random sampling is used for this quantitative study. Additionally, sampling technique relies on the kind of data researchers are looking for. The attempt to reduce bias is done. In terms of sampling bias, this case did not regard generalized outcomes.

The term “Reliability” of a data set implies whether in sample, variables are stationary or not. It may be implied that the researcher can repeat their task by another researcher to obtain similar outcomes (Winter, 2000). The study is found therefore reliable. This research is valid as the analysis produced the intended outcome. It is known that simple random sampling provides a representative sample for the total population. Therefore, the researcher could have make the research more practical by using simple random sampling. The other variables especially continuous variables such as intelligence level may additionally predict the dependent variable more nicely.

The survey questionnaire contains total 16 questions. The primary part of the survey involves different types of demographic information such as place, age, gender and nationality. The next 12 questions give the information about both dependent and independent variables. Five control variables put an impact on both the dependent and independent variables. The current research contains continuous, ordinal, nominal and dummy variables for hypothesis-testing. To assess the Academic Performance, the responses based on questionnaire perceive themselves to fit. Most of the variables are categorical data (nominal data and ordinal data). “Age”, “Hours of lecture attended” and “Average Working Hours” are the numerical variables of the data set. These variables are continuous in nature. The categorical variables are firstly labelled as numerical values. Then, these are transformed into dummy variables (levels are only 0 and 1). These variables helped to test the hypothesis. The rejection of acceptance of null hypothesis is decided as per three multiple regression models. The statistical software “R” is used for statistically analysing the data. The codes and variables of the research are as following:

Variable Name	Coding	Type of Variable
Gender	Male (1), Female (2)	Dummy variable
Nationality	Korea (1), Norway (2), USA (3), Uganda (4), Persia (5), Canada (6), Lithuania (7), Ethiopia (8), UK (9), Israel (10), China (11), Hong Kong (12), Ghana (13), Sweden (14),	Nominal variable
Study Level	Bachelor (1), Masters (2), PHD (3)	Ordinal variable
Hours of Lecture per week	2-6 hours (1), 6-10 (2), 10-14 (3)	Ordinal variable
Grade	A (1), B (2), C (3), D (4), E (5), F (6)	Ordinal variable
Part Time Job	Yes (1), No (2)	Dummy variable
Work Clash with Work	Yes (1), No (2)	Dummy variable
Academic Performance	Strongly Disagree (1), Disagree (2), Indifferent (3), Agree (4), Strongly Agree (5)	Ordinal variable

The researcher transferred the data collected from “Google forms” to excel sheet in .csv or .xlsx format. In this data set, the missing value of “Gender” is cleaned up. No other variable is cleaned for multiple regression models. Before analysis, the researcher coded the data into numbers for making easier analysis. Additionally, control variables (independent variables) are other variables that impacts academic performance of the students. The data involves the variable “Gender” that has two labels males and females. The control variables of the research report are “Gender”, “Nationality”, “Study level”, “Hours of Lecture per week”, “Grade” and “Work Clash with Work”.

	Min	1st Quartile	Median	Mean	3rd Quartile	Max	Missing	Total
Gender	0.000	0.000	2.000	0.689	1.000	1.000	1	41
Nationality	1.000	2.000	7.500	8.524	3.000	21.000	0	42
Study Level	1.000	2.000	2.000	1.786	2.000	2.000	0	42
Hours of Lecture per week	1.000	2.000	2.000	2.238	3.000	3.000	0	42
Grade	1.000	2.000	2.000	2.286	3.000	3.000	0	42
Part Time Job	0.000	0.000	0.000	0.333	1.000	1.000	0	42
Work Clash with Work	0.000	0.000	1.000	0.657	1.000	1.000	7	35
Academic Performance	1.000	2.250	4.000	3.476	4.000	5.000	0	42

Data Analysis

The table of descriptive statistics indicates the crucial components of collected data. It indicates location measures, central tendencies and dispersion measures along with number of variables and name of the variables of the data set (Oja, 1983).

The analysis section represents graphical description of the data and elaborates the outcomes of multiple regression analysis of three models. The descriptive summary of “Part Time Job” and “Academic Performance” is presented by histogram plots in the below section.

Histogram drawing is the best way to present the frequency distribution of the numerical variables. Histogram gives exact information about the distribution of the variable. It helps to find out the normality of the distribution. Besides the it helps to find out skewness and “modal” values. This is crucial as it helps in focusing the most relevant portion of the distribution.

Figure 1: The histogram shows the association between the dependent variables and its frequency

(Pizer et al., 1987)

The figure refers that the distribution of academic performance is normally distributed. The distribution is irregular in shape. The distribution is negatively skewed that means that median is greater than mean. It proves that most of the respondents have performed well enough. Most of the people agree that their academic performance is good whereas some of them agree that their academic performance is not good. The academic performance is measured in “Likert” scale where 1 to 5 is denoted by strongly disagree, disagree, indifferent, agree and strongly agree respectively. Approximately 2% frequency of the distribution referred that students utterly disagreed with the statement.

Figure 2: The histogram displays the association between the independent variables and its frequency

This histogram refers the frequency distribution of the independent variable “Part time job”. The figures refer that the distribution is not normal. The distribution is also found to be positively distributed. It could be inferred from the histogram that the number of people who do part time job is greater in number than the number of people who are not involved in part time job. The variable has dichotomous responses. From the histogram, it is clear that the frequency of people who have part time job are almost double in frequency who perform part time job. Note that, the researcher avoided the presence of frequencies of missing data.

Table 1: Analysis of Multiple Regression Model

Variables / Dependent Variable: Academic Performance	Coef (Std Error) Model 1	Coef (Std Error) Model 2	Coef (Std Error) Model 3
Gender		-0.8886 (0.3899)	-1.0107 (0.4220)
Nationality		0.0279 (0.0275)	0.0293 (0.0349)
Study Level		0.4767 (0.4113)	0.3566 (0.4649)
Hours of Lecture per week		0.0845 (0.2312)	0.2443 (0.2834)
Grade			0.2475 (0.3372)
Part Time Job	0.250 (0.366)	0.6326 (0.3852)	0.7622 (0.5252)
Work Clash with Work			-0.4587 (0.4516)
Academic Performance
Intercept	3.393 (0.211)	2.5737 (0.9413)	2.1671 (1.3700)
Number of Observations	42	37	28
R2	0.0116	0.239	0.308
p-value	0.498	0.076	0.165

(Aiken, West & Reno, 1991)

Model 1: Dependent variable = Academic Performance

Data Collection Process

Independent variable = Part Time Job.

Model 2: Dependent variable = Academic Performance

Independent variable = Gender, Nationality, Study level, Hours of Lecture per week, Part Time Job,

Model 3: Dependent variable = Academic Performance

Independent variable = Gender, Nationality, Study level, Hours of Lecture per week, Grade, Part Time Job and Work Clash with Work.

Model 1 does not consist control variables. However, Model 2 and Model 3 include control variables that explains the variation that exists between “Academic Performance” and “Part time Job”. The control variables of Model 2 are Gender, Nationality, Study level and Hours of Lecture per week. The control variables of Model 3 are Gender, Nationality, Study level, Grade, Hours of Lecture per week and Work Clash with Work. The Multiple regression models could give more information about the influence of independent variables on the association between dependent variable and independent variable (Mason & Perreault, 1991). The coefficient of multivariate analysis are actually the slopes of the different variables towards the fitted regression model (Cohen et al., 2013).

For the Model 1, the value of R² is 0.0116. Therefore, independent variable (Part Time Job) can explain only 1.16% variability of the dependent variable (Academic Performance). The low value of “Coefficient of determination” (R²) interprets that the fitting of the model is not very good (Draper & Smith, 2014). The correlation of dependent and independent variable is also negligible. The slope refers that the correlation between dependent and independent variable is positive.

Model 2 provides much better interpretation due to the presence of control variables. The slopes of the predictor of control variables indicate that-

• Gender has insignificant negative association with Academic performance (coefficient = -0.8886, p-value = 0.3899) (Smouse, Long & Sokal, 1986).
• Nationality has significant positive association with Academic performance (coefficient = 0279, p-value = 0.0275).
• Study level has insignificant positive association with Academic performance (coefficient = 4767, p-value = 0.4113).
• Hours of lecture per week has insignificant positive association with Academic performance (coefficient = 0845, p-value = 0.2312).
• Part time job has positive insignificant relevance with Academic performance (coefficient = 6326, p-value = 0.3852).

• Gender has insignificant negative association with Academic Performance (slope = -1.0107 and p-value = 0.4220).
• Nationality has significant positive association with Academic Performance (slope = 0293 and p-value = 0.0349).
• Study level has insignificant positive association with Academic Performance (slope = 3566 and p-value = 0.4649).
• Hours of Lecture per week has insignificant positive association with Academic Performance (slope = 2443 and p-value = 0.2834).
• Grade has insignificant positive association with Academic Performance (slope = 0.2475 and p-value = 0.3372).
• Part Time Job has insignificant positive association with Academic Performance (slope = 0.7622 and p-value = 0.5252).
• Work Clash with Work has insignificant positive association with Academic Performance (slope = -0.4587 and p-value = 0.4516).

The value of multiple R² is 0.308. Therefore, 30.8% of the dependent variable (Academic Performance) is explained by Part time job (independent variable), when all other variables are treated as control variables.  

It is a notable fact that in Model 3, the inclusion of dummy and control variables has developed the goodness of fitting of the model. In the Model 3, two extra variables are added as Grades and Work Clash with Work. The value of coefficient of determination is greater for Model 3 than Model 2 and Model 1. The model 3 has significant p-value 0.165, whereas the model 2 has significant p-value 0.076 and model 1 has significant p-value 0.498. The p-value is least for model 2; however no model has significant p-value less than 0.05. It could be interpreted that no model could prove the hypothesis that part time work is significantly associated with the dependent variable Academic performance (Source: “RPubs – Multiple Linear Regression in R – First Steps”, 2018).

• Gender, Study level, Hours of Lecture per week and part time jobare insignificant control variables in both Model 2 and Model 3.
• Nationality is significant control variables in both Model 2 and Model 3.
• In Model 3, Gender and Study level are less correlated than Model 2.
• In Model 3, Nationality and Hours of Lecture per week are more correlated than Model 3.
• The correlation of Part time job with Academic performance is highest for Model 3 followed by Model 2. The correlation is least for Model 1 (Hair et al., 1998).
• The value of intercept has decreased gradually from Model 1 to Model 2 and finally Model 3.

Sampling Strategy

Therefore, the Model 3 is best among the entire three models. Overall, the model 3 provides maximum information among all the models. The results are drawn as per 5% level of significance. Model 3 gives more information about the relationship than simple model (Model 1) and improved model (Model 3). The control-variables are therefore are proved to be very helpful.

P-value has decreased for Model 2 than Model 1. It refers the improvement of the model (Aiken, West & Reno, 1991). However, the p-value again has increased for Model 3. P-value of the multiple regression models interprets that Model 2 is best. Where p-value signifies the association between dependent and independent variables, then R² indicates the validity and fitness of the model. Therefore, considering all the Models, Model 3 is found to be most effective model to establish the research objective.

In all the models, null hypothesis is accepted at 5% level of significance and alternative hypothesis is accepted (Jaccard, Wan & Turrisi, 1990). In all the three models, it could be stated that Part Time Job has no statistical significance to the response variable Academic performance.

The researcher crosschecked the calculation of the analysis in SPSS-20. It ensures the validity and quality of the outcomes. The research took lots of time to collect data, decide the hypothesis and analyze the dataset in R. The understanding and interpretation of the regression models helped to detect the biases in the outcomes.

Conclusion:

The quantitative research study was utilised to detect the association between Part time job and Academic performance. The data analysis was executed as per responses of the interviewers. In data analysis, the relevance of dependent and independent variables are assessed to find the correlation and association. The outcomes from the analysis refers that there is no significant correlation between IV (independent variable) and DV (dependent variable). The more inclusion of independent and control variables would make the model better. With the increment or decrease of IV, the DV also increases insignificantly. Simultaneously, with the decrease of IV, the DV also increases insignificantly. These models indicate the positive relationships between DV and all the variables except gender and work clashes with work. The values of multiple R² in these models are very low. Neither bivariate nor the multivariate models are statistically significant. The model as upgraded to model 2 and model 3, it is observed that the model started to fit better. More control variables might provide better results. More variables are needed to apply in this model as the value of R² is gradually increasing. It suggests that variation is far better being explained by recent model. The p-value suddenly decreased from model 2 to model 3. Hence, it is important to select appropriate dummy variables for analysis.

Reliability and Validity of the Study

In conclusion, the alternative hypothesis is accepted for regression models at 5% level of significance. All the models have p-value greater than 0.05. It refers that significance and p-value provides evidences to reject the null hypothesis. Additionally, both positive and negative correlation was produced from regression models 2 and 3. However, the value of R² is too small for considering the significance of the relationship. The high p-values displays that the association between DV and ID s are statistically insignificant. It supports the null hypothesis that there is no association between academic performance and part time jobs. Even while control variables were added, the outcomes showed same trend. These imply that control variables have insignificant impact on academic performance. Generally, model 3 provided better result in the research study.

The limitation of the study is that some samples got omitted from model 1 to model 2 and finally model 3. If the more samples would be present and no missing values would be occurred, then it would have explained the dependent variable better. The limitation is sampling technique is also utilised in this model. Used covariance sampling represent the population and therefore outcomes for this study must not be generalised for all the students. Covariance-sampling method introduces bias excluding other students. Sample number of samples used in this analysis causes high standard error values referring the deviation of a sample mean from the population. In model 3, the standard error has increased from model 2. Greater standard weakens the strength of the model. The outcomes make representative of the population weak. The other limiting data with this analysis is missing data. It has contributed low R², higher p-value and higher standard error.

References:

Aiken, L. S., West, S. G., & Reno, R. R. (1991). Multiple regression: Testing and interpreting interactions. Management.

Aiken, L. S., West, S. G., & Reno, R. R. (1991). Multiple regression: Testing and interpreting interactions. Sage.

Bryman, A. (2015). Social research methods. Oxford university press.

Bryman, A. (2017). Quantitative and qualitative research: further reflections on their integration. In Mixing methods: Qualitative and quantitative research (pp. 57-78). Routledge.

Cohen, J., Cohen, P., West, S. G., & Aiken, L. S. (2013). Applied multiple regression/correlation analysis for the behavioral sciences. Routledge.

Draper, N. R., & Smith, H. (2014). Applied regression analysis(Vol. 326). John Wiley & Sons.

Eberly, L.E., 2007. Multiple linear regression. In Topics in Biostatistics. Humana Press. (pp. 165-187).

Farley, J. U., & Hinich, M. J. (1970). A test for a shifting slope coefficient in a linear model. Journal of the American Statistical Association, 65(331), 1320-1329.

Grégoire, G., 2014. Multiple linear regression. European Astronomical Society Publications Series, 66, pp.45-72.

Hair, J. F., Black, W. C., Babin, B. J., Anderson, R. E., & Tatham, R. L. (1998). Multivariate data analysis (Vol. 5, No. 3, pp. 207-219). Upper Saddle River, NJ: Prentice hall.

Jaccard, J., Wan, C. K., & Turrisi, R. (1990). The detection and interpretation of interaction effects between continuous variables in multiple regression. Multivariate behavioral research, 25(4), 467-478.

Koch, K. R. (2013). Parameter estimation and hypothesis testing in linear models. Springer Science & Business Media.

Luthans, B. C., Luthans, K. W., & Jensen, S. M. (2012). The impact of business school students’ psychological capital on academic performance. Journal of Education for Business, 87(5), 253-259.

Macan, T. H. (1996). Time-management training: Effects on time behaviors, attitudes, and job performance. The Journal of psychology, 130(3), 229-236.

Macan, T. H., Shahani, C., Dipboye, R. L., & Phillips, A. P. (1990). College students’ time management: Correlations with academic performance and stress. Journal of educational psychology, 82(4), 760.

Mason, C. H., & Perreault Jr, W. D. (1991). Collinearity, power, and interpretation of multiple regression analysis. Journal of marketing research, 268-280.

Murphy, K. R., Myors, B., & Wolach, A. (2014). Statistical power analysis: A simple and general model for traditional and modern hypothesis tests. Routledge.

Oja, H. (1983). Descriptive statistics for multivariate distributions. Statistics & Probability Letters, 1(6), 327-332.

Part-Time Work and Student Achievement – Educational Leadership. Retrieved from https://www.ascd.org/publications/educational_leadership/apr01/vol58/num07/_Part-Time_Work_and_Student_Achievement.aspx

Pizer, S. M., Amburn, E. P., Austin, J. D., Cromartie, R., Geselowitz, A., Greer, T., … & Zuiderveld, K. (1987). Adaptive histogram equalization and its variations. Computer vision, graphics, and image processing, 39(3), 355-368.

Robotham, D. (2012). Student part-time employment: characteristics and consequences. Education+ Training, 54(1), 65-75.

RPubs – Multiple Linear Regression in R – First Steps. (2018). Rpubs.com. Retrieved 3 April 2018, from https://rpubs.com/FelipeRego/MultipleLinearRegressionInRFirstSteps

Smouse, P. E., Long, J. C., & Sokal, R. R. (1986). Multiple regression and correlation extensions of the Mantel test of matrix correspondence. Systematic zoology, 35(4), 627-632.

The effects of part-time work on school students. (2018). Research.acer.edu.au. Retrieved 3 April 2018, from https://research.acer.edu.au/cgi/viewcontent.cgi?article=1017&context=lsay_research

Wehrens, R., & Mevik, B. H. (2007). The pls package: principal component and partial least squares regression in R.

Winter, G. (2000). A comparative discussion of the notion of’validity’in qualitative and quantitative research. The qualitative report, 4(3), 1-14.