Skip to main content

Advertisement

SpringerLink
Log in

Does Work Harm Academic Performance of Students? Evidence Using Propensity Score Matching

  • Published:
Research in Higher Education Aims and scope Submit manuscript

Abstract

In this article we analyze the effects of student work on academic performance for college students. In order to reduce the endogeneity bias due to selection into treatment, we use propensity score matching technique. This approach allows us to estimate the effects of student work separately for different years of study, which is not possible when inside instruments are used to deal with endogeneity of student work. We find predominantly negative effects of student work for all measures of academic performance (GPA, exam attempts, exams passed, and likelihood of passing a year), although many of these are economically and statistically insignificant. We supplement existing studies that do not estimate separate treatment effects for different years of study by showing that work while in college harms study outcomes mostly in the first year of study—by passing smaller number of exams and thereby increasing the likelihood of failing a year. Our results are consistent with evidence on difficulty with adjusting to college studies of first-year students, who face many uncertainties that affect finding the optimal allocation of time between studies, work and leisure.

This is a preview of subscription content, log in via an institution to check access.

Access this article

Log in via an institution

Price excludes VAT (USA)
Tax calculation will be finalised during checkout.

Instant access to the full article PDF.

Institutional subscriptions

Similar content being viewed by others

Notes

  1. We find evidence of small positive effects of student work experience on wages and likelihood of first regular employment, especially when (i) jobs can be deemed as high skilled and (ii) related to the field of studies.

  2. Ehrenberg and Sherman (1987) found that only off-campus work negatively affected graduation-on-time and drop-out probabilities in the third and fourth year of study.

  3. These amounts of student work are comparable to those used by other studies. For example, Beerkens et al. (2011) use three treatment intervals with boundaries set at 9 h per week and 25 h per week.

  4. The high-school grades range between 1 (insufficient) and 5 (excellent), and 2 is the lowest passing grade. The matura exam consists of three compulsory (Slovene language, Mathematics, and one foreign language—usually English) and two elective subjects, such as Biology, History, Physics, etc.).

  5. These grades were often a simple mapping from achieved percentages like \(Grade = int(Score/10)+1\), where int denotes the integer part of the ratio.

  6. A useful summary of Slovenian labor markets during the period of analysis is given in OECD (2009).

  7. We are grateful to SORS for all assistance in creating our final data set.

  8. Student finishing vocational matura had to pass additional subject of general matura in order to be able to enrol to university courses.

  9. Although the negative grades range from 1 to 5, grades 1 to 4 were rarely used by some examiners and instead 5 was given to all students who did not achieve the required minimum threshold to pass an exam.

  10. We assume the average number of working hours per month is 175, which is equivalent to 2100 working hours per year.

  11. We count as family members parents and children under the age of 27, following the personal-income-tax act that defines as a dependent family member a person up to the age of 26 (in addition to other requirements).

  12. Conditional on x, outcomes of treatment (\(y_1\)) and control group (\(y_0\)) are independent of treatment (D). Rosenbaum and Rubin (1983) showed that if the former holds, \(y_1\) and \(y_0\) are also independent of D for given value of propensity score.

  13. For every value of propensity score, there are observations in both control and treatment groups.

  14. We also considered other matching algorithms and other caliper values but obtained qualitatively similar estimated effects. We chose this method because it is in line with the recommendation to make the control group as locally comparable as possible to the treated, and baseline differences as little as possible in order to estimate the treatment effects using comparable subjects (Lee 2005).

  15. We also considered alternative intervals for calculation of estimated effects (e.g. 2-month intervals), but also faced violation of the balancing property in some cases. When balancing property was satisfied, the results we present and those for smaller intervals were nevertheless comparable.

  16. The set of controls increases with years of study as we include past academic performance measures from the second year onwards and indicators for selected majors from the third year of study.

  17. Note that the total number of attempts to exams does not have an a priori negative relationship with student work. Students who work more are less likely to pass and consequently may exhibit more attempts to pass exams.

  18. Note that this pattern is not unique to our data. Darolia (2014) finds a similar inverse U-shaped pattern using US National Longitudinal Youth Survey 1997 with peak around 5 h of work per week for full-time students, which is equivalent to 1.5 months of work per year.

  19. For example, a fee could amount to 5% of transferred cash.

  20. The number of unbalanced variables before matching is between 3 and 17, varying with the effect of interest and year of study.

  21. The number of unbalanced variables ranged between 0 and 4, which still presents a significant improvement in the balancing property of the matched data set compared to the unmatched one.

  22. For example, the estimated effect of student work on number of exams passed when increasing work from less than 2 months to more than 7 months is −1.00 based on actual hours and −1.28 based on real pay.

References

  • Auers, D., Rostoks, T., & Smith, K. (2007). Flipping burgers or flipping pages? Student employment and academic attainment in post-soviet latvia. Communist and Post-Communist Studies, 40(4), 477–491.

    Article  Google Scholar 

  • Austin, P. C. (2011). Optimal caliper widths for propensity-score matching when estimating differences in means and differences in proportions in observational studies. Pharmaceutical Statistics, 10(2), 150–161.

    Article  Google Scholar 

  • Baker, R. W., & Siryk, B. (1984). Measuring adjustment to college. Journal of Counseling Psychology, 31, 1251–1271.

    Google Scholar 

  • Bartolj, T., & Polanec, S. (2016). The game of skills: A comparison of impacts of student work and academic performance on post-college labour market outcomes. Mimeo

  • Beerkens, M., Mägi, E., & Lill, L. (2011). University studies as a side job: Causes and consequences of massive student employment in Estonia. Higher Education, 61(6), 679–692.

    Article  Google Scholar 

  • Callender, C. (2008). The impact of term-time employment on higher education students’ academic attainment and achievement. Journal of Education Policy, 23(4), 359–377.

    Article  Google Scholar 

  • Darolia, R. (2014). Working (and studying) day and night: Heterogeneous effects of working on the academic performance of full-time and part-time students. Economics of Education Review, 38, 38–50.

    Article  Google Scholar 

  • Dehejia, R. H., & Wahba, S. (2002). Propensity score-matching methods for nonexperimental causal studies. Review of Economics and Statistics, 84(1), 151–161.

    Article  Google Scholar 

  • DeSimone, J. (2006). Academic performance and part-time employment among high school seniors. Topics in Economic Analysis & Policy, 6(1), 13.

    Article  Google Scholar 

  • DeSimone, J. S. (2008). The impact of employment during school on college student academic performance. National Bureau of Economic Research: Technical reports

  • Dustmann, C., & Soest, A. (2007). Part-time work, school success and school leaving. Empirical Economics, 32(2–3), 2–3.

    Google Scholar 

  • Eckstein, Z., & Wolpin, K. I. (1999). Why youths drop out of high school: The impact of preferences, opportunities, and abilities. Econometrica, 67(6), 1295–1339.

    Article  Google Scholar 

  • Ehrenberg, R. G., & Sherman, D. R. (1987). Employment while in college, academic achievement, and postcollege outcomes: A summary of results. Journal of Human Resources, 22(1), 1–23.

    Article  Google Scholar 

  • Hirano, K., & Imbens, G. W. (2004). The propensity score with continuous treatments. In A. Gelman & X. L. Meng (Eds.), Applied bayesian modeling and causal inference from incomplete-data perspectives (pp. 73–84). Chichester: Wiley.

    Google Scholar 

  • Kahn, J. H., & Nauta, M. M. (2001). Socialcognitive predictors of first-year college persistence: The importance of proximal assessment. Research in Higher Education, 42(6), 633–652.

    Article  Google Scholar 

  • Kalenkoski, C. M., & Pabilonia, S. W. (2010). Parental transfers, student achievement, and the labor supply of college students. Journal of Population Economics, 23(2), 469–496.

    Article  Google Scholar 

  • Kalenkoski, C. M., & Pabilonia, S. W. (2012). Time to work or time to play: The effect of student employment on homework, sleep, and screen time. Labour Economics, 19(2), 211–221.

    Article  Google Scholar 

  • Lee, C., & Orazem, P. F. (2010). High school employment, school performance, and college entry. Economics of Education Review, 29(1), 29–39.

    Article  Google Scholar 

  • Lee, M. J. (2005). Micro-econometrics for policy, program, and treatment effects. Oxford: Oxford University Press.

    Book  Google Scholar 

  • Lillydahl, J. H. (1990). Academic achievement and part-time employment of high school students. Journal of Economic Education, 21(3), 307–316.

    Article  Google Scholar 

  • OECD. (2009). Reviews of labour market and social policies for slovenia. Technical report. Paris: OECD Publishing.

  • Oettinger, G. S. (1999). Does high school employment affect high school academic performance? Industrial and Labor Relations Review, 53, 136–151.

    Article  Google Scholar 

  • Pascarella, E. T., & Terenzini, P. (1983). Predicting voluntary freshman year persistence/withdrawal behavior in a residential university: A path analytic validation of tinto’s model. Journal of Educational Psychology, 75(2), 215–226.

    Article  Google Scholar 

  • Post, D., & Pong, S. L. (2009). Student labour and academic proficiency in international perspective. International Labour Review, 148(1–2), 93–122.

    Article  Google Scholar 

  • Quirk, K. J., Keith, T. Z., & Quirk, J. T. (2001). Employment during high school and student achievement: Longitudinal analysis of national data. The Journal of Educational Research, 95(1), 4–10.

    Article  Google Scholar 

  • Rosenbaum, P. R., & Rubin, D. B. (1983). The central role of the propensity score in observational studies for causal effects. Biometrika, 70(1), 41–55.

    Article  Google Scholar 

  • Rothstein, D. S. (2007). High school employment and youths’ academic achievement. Journal of Human Resources, 42(1), 194–213.

    Article  Google Scholar 

  • Ruhm, C. J. (1997). Is high school employment consumption or investment? Journal of Labor Economics, 15(4), 735–776.

    Article  Google Scholar 

  • Singh, K. (1998). Part-time employment in high school and its effect on academic achievement. The Journal of Educational Research, 91(3), 131–139.

    Article  Google Scholar 

  • SSC Act. (2001) Social security contributions act. Official Gazette of the RS [Zakon o prispevkih za socialno varnost, Uradni list RS] no. 5/1996, 34/1996, 97/2001

  • Stinebrickner, R., & Stinebrickner, T. R. (2003). Working during school and academic performance. Journal of Labor Economics, 21(2), 473–491.

    Article  Google Scholar 

  • Tyler, J. H. (2003). Using state child labor laws to identify the effect of school-year work on high school achievement. Journal of Labor Economics, 21(2), 381–408.

    Article  Google Scholar 

Download references

Author information

Authors and Affiliations

  1. Institute for Economic Research, Kardeljeva pl. 17, 1000, Ljubljana, Slovenia

    Tjaša Bartolj

  2. Faculty of Economics, University of Ljubljana, Kardeljeva pl. 17, 1000, Ljubljana, Slovenia

    Sašo Polanec

  3. The Bank of Slovenia, Slovenska 35, 1505, Ljubljana, Slovenia

    Sašo Polanec

Authors
  1. Tjaša Bartolj
    View author publications

    You can also search for this author in PubMed Google Scholar

  2. Sašo Polanec
    View author publications

    You can also search for this author in PubMed Google Scholar

Corresponding author

Correspondence to Tjaša Bartolj.

Ethics declarations

Conflict of interest

The authors declare that they have no conflict of interest.

Appendix

Appendix

Tables 7, 8, 9, 10 and 11 and Figs. 1, 2, 3, 4, 5 and 6.

Table 7 Structure of full sample, by year of study and region
Full size table
Table 8 Full sample size, by amount of student work
Full size table
Table 9 eŠtudentski servis sample size, by study year and gender
Full size table
Table 10 Summary statistics for eŠtudentskis servis sample
Full size table
Table 11 Structure of eŠtudentski servis sample, by region
Full size table
Fig. 1
figure 1

Representation of causal chain

Full size image
Fig. 2
figure 2

Academic performance by student work in the first year of study. The size of markers is proportional to the frequency of students with a specific value of student work. Markers with frequency lower than 10 are omitted. The number of months worked is calculated by dividing the nominal income earned by the average wage rate (for all students) and the average number of hours per month

Full size image
Fig. 3
figure 3

Academic performance by student work in the second year of study. The size of markers is proportional to the frequency of students with a specific value of student work. Markers with frequency lower than 10 are omitted. The number of months worked is calculated by dividing the nominal income earned by the average wage rate (for all students) and the average number of hours per month

Full size image
Fig. 4
figure 4

Academic performance by student work in the third year of study. The size of markers is proportional to the frequency of students with a specific value of student work. Markers with frequency lower than 10 are omitted. The number of months worked is calculated by dividing the nominal income earned by the average wage rate (for all students) and the average number of hours per month

Full size image
Fig. 5
figure 5

Academic performance by student work in the fourth year of study. The size of markers is proportional to the frequency of students with a specific value of student work. Markers with frequency lower than 10 are omitted. The number of months worked is calculated by dividing the nominal income earned by the average wage rate (for all students) and the average number of hours per month

Full size image
Fig. 6
figure 6

Student work income distribution by year. The histograms show the student work income distributions for each study year and the sum of general tax deduction and study-specific tax deduction (vertical line)

Full size image

Rights and permissions

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Bartolj, T., Polanec, S. Does Work Harm Academic Performance of Students? Evidence Using Propensity Score Matching. Res High Educ 59, 401–429 (2018). https://doi.org/10.1007/s11162-017-9472-0

Download citation

  • Received:

  • Published:

  • Issue Date:

  • DOI: https://doi.org/10.1007/s11162-017-9472-0

Keywords

Search

Navigation