Internships can improve graduate career outcomes, but the effect is smaller and more conditional than raw comparisons suggest. Studies that compare interns with non-interns mostly capture selection: motivated, well-connected or high-achieving students are more likely to take internships. Research designs that address selection, such as instrumental-variable studies in Germany and Switzerland, a randomized program in Yemen and employer vignette experiments in Belgium, still find positive effects on earnings, employment or hiring chances. The evidence points to design as the deciding factor: structured internships with a mentor, adequate duration and supervision that supports learning are associated with better outcomes than internships that are simply completed.
Why a simple comparison of interns and non-interns misleads
Almost every graduate survey shows that former interns do better than graduates without an internship. The difficulty is that the two groups were not alike before the internship began. Students who secure internships may have higher grades, clearer career plans, stronger networks or more money to accept low-paid work. Each of these traits can improve later outcomes on its own. When a university reports that its interns earn more, it cannot tell from that number alone how much of the gap the internship caused.
A literature review commissioned by the Department for Education in England found exactly this problem in UK research published since 2010. The reviewers, from the Warwick Institute for Employment Research, report that placements were generally associated with positive outcomes, that the majority of studies did not address self-selection, and that the few studies that did found a lower, but still positive, association.
One UK study shows how large the selection gap can be. Jones, Green and Higson (2017) used student records from two UK universities (6,645 students at one, 737 business students at the other), where every student could choose whether to take a sandwich placement year. A simple comparison showed that placement students scored 5.3 and 5.6 percentage points higher in final-year marks. After propensity score matching on observed characteristics, the estimated placement effect fell to 2.6 and about 3.8 points. Roughly half of the raw advantage in one institution, and about a third in the other, reflected who chose to go on placement rather than what the placement did. The outcome here is academic performance, not employment, but the lesson about selection carries over.
What each kind of study can and cannot show
The research on internships uses designs of very different strength. Readers comparing claims need to know which design sits behind a number.
| Design | What it can show | Main weakness | Example discussed here |
|---|---|---|---|
| Raw comparison of interns and non-interns | Whether outcomes differ between the groups | Mixes the internship effect with selection into internships | Most institutional destination reports |
| Regression or matching on observed traits | Whether a gap remains after accounting for measured differences such as prior grades | Cannot rule out unmeasured traits such as motivation or networks | Jones et al. (2017); Di Meglio et al. (2022); Bittmann and Zorn (2020) |
| Instrumental variables based on internship requirements | The effect for students who took an internship because their program required it | Relies on the requirement affecting outcomes only through the internship; estimates apply to that group of students | Margaryan et al. (2022); Bolli et al. (2021) |
| Randomized program | The causal effect of being offered an internship in that program | Results may not transfer to other countries, sectors or study-based internships | McKenzie et al. (2016) |
| Vignette or audit experiments with employers | How an internship on a CV changes interview or hiring decisions | Measures employer reactions, not later careers or skills | Tobback et al. (2024) |
| Surveys of perceived skills and satisfaction | Which features students or graduates associate with learning | Self-reported, and usually no comparison group | Hora et al. (2024); Jackson and Dean (2023) |
What the stronger evidence finds
Earnings: instrumental-variable studies
Two studies use the fact that some degree programs require internships and others do not. Margaryan, Saniter, Schumann and Siedler (2022) used mandatory firm internships at German universities as an instrument and longitudinal graduate survey data. According to the abstract, firm internships were associated with earnings returns of about 6 percent in both ordinary least squares and instrumental-variable estimates, with larger returns in fields with weak labor market orientation, and a lower risk of unemployment in the first year of work.
Bolli, Caves and Oswald-Egg (2021) applied a similar strategy to 8,615 graduates of Swiss conventional universities from three waves of the Swiss Graduate Survey run by the Federal Statistical Office, measuring income one and five years after graduation. Their instrumental-variable estimates imply a 17.0 percent short-term and 14.9 percent long-term wage gain from an internship. The authors themselves caution that instrumental-variable approaches tend to yield high coefficients and that they prefer to focus on interpretation rather than effect size. They also note that the survey's response rates ranged from about 30 to 60 percent across waves. Their interpretation is that internships build general workplace skills rather than firm-specific skills, because graduates who later switched employers or fields also gained.
Hiring decisions: employer experiments
Tobback, Verhaest and Baert (2024) asked human resources professionals in Belgium to evaluate fictitious candidates that differed only in their internship history. Internships improved interview and hiring chances, and voluntary extra-curricular internships had the strongest effect, compared with mandatory or voluntary internships inside the curriculum. Employers read internships as evidence of skills acquired during study and of pre-existing ability and motivation. Bolli and colleagues summarize two earlier résumé audit studies as reporting higher interview rates for applicants who listed an internship; those two studies are not cited directly here because their publisher pages could not be verified for this article.
A randomized graduate internship program
The only randomized evidence located for this article comes from a program for recent university and vocational graduates in Yemen. McKenzie, Assaf and Cusolito (2016) report that 500 applicants were randomly assigned within strata of specialization and city; the program paid half the cost of six-month internships and matched graduates with firms. Receiving an internship raised the likelihood of working during the internship period by 42 percentage points and increased income by 73 percent, and recipients had better employment outcomes in the first five months after the program. The authors note that they could not assess whether these gains came at the expense of other job seekers, and the follow-up survey was fielded as civil conflict was beginning, which also led to the cancellation of the program's second year. This is a post-graduation program in a fragile economy, so it shows that internship experience can have causal effects, not that a university's own internships do.
Medium-term employment and job quality
Di Meglio, Barge-Gil, Camiña and Moreno (2022) used Spain's first University Graduate Job Placement Survey (2014) and tested how sensitive their estimates were to omitted variables. Internship experience eased the transition to the first job and its link with employment persisted four years later, but evidence of better job matching was weak and no effects on wages were found. Taken together with the German and Swiss studies, the pattern is that employment-entry benefits are the most consistent finding, while long-term wage effects vary by country and method.
Which design features appear to matter
Voluntary, required or embedded
Three studies in different countries point in the same direction. Bittmann and Zorn (2020) found, using Austrian graduate data and regression models, that voluntary internships were associated with better income, less job mismatch and higher job satisfaction, while mandatory internships showed no comparable association; the association grew linearly with the length of voluntary internships. Tobback and colleagues found the strongest employer response to voluntary extra-curricular internships. Jackson and Bridgstock (2021), surveying 510 business and creative industries graduates from three Australian universities, found that graduates rated extra-curricular internships more highly than internships delivered as work-integrated learning.
These findings do not show that requirements are useless. The German and Swiss instrumental-variable studies estimate positive effects precisely for students who took internships because their program required them. A plausible reading is that voluntary internships carry an extra signal of motivation that employers reward, while required internships still deliver real learning for students who would not otherwise have taken one.
Structure, mentoring, duration and pay
An International Labour Organization working paper by O'Higgins and Pinedo (2018) analyzed the 2013 Flash Eurobarometer survey of traineeships, which randomly sampled 13,422 people aged 18 to 35 across the EU28; 5,484 of them reported at least one internship, and the main analysis focused on former interns who had finished their studies. The paper focuses on internships in general, including open-market internships after education, not only internships within degree programs. Its probit models show that structured internships, with a mentor, health insurance, a completion certificate and working conditions similar to regular staff, were associated with learning useful things and with perceived usefulness for finding work. Longer internships and larger firms were also associated with learning. Pay was not associated with learning, but paid interns were more likely to be offered an extension or a job by the host firm. In the medium term, having a mentor and a certified internship remained associated with current employment, while pay did not. The authors state that the survey was not designed as an impact evaluation and that the coefficients reflect associations.
Supervision and task design
Hora, Chen, Wolfgram, Zhang and Fischer (2024) combined surveys of 435 students and focus groups with 52 students at five institutions. Supervisor behaviors were significantly associated with intern satisfaction and perceived development. In the focus groups, clear communication, availability and feedback marked positive supervision, while unavailability and inattention to learning marked negative experiences. The authors recommend designing internships as mentored learning spaces, with attention to task scaffolding and student autonomy, and training supervisors.
Workplace and non-workplace learning
Jackson and Dean (2023) analyzed responses from 76,261 graduates in Australia's national Graduate Outcomes Survey. Work-based placements were associated with perceived skill gains, but non-workplace work-integrated learning was associated with larger perceived skill gains, particularly for collaborative skills and among postgraduates. Work-based placements were most strongly associated with graduates' sense of overall preparedness for employment. These are self-reported perceptions rather than labor market outcomes. Simulation-based learning, one non-workplace approach, is covered in an earlier UNIRANKS article on experiential simulation learning.
Our interpretation
Read together, the studies support four conclusions. First, internships can cause better early career outcomes; this is shown by designs that address selection, not only by raw comparisons. Second, the size of the benefit is uncertain and context-dependent: estimates range from no wage effect in Spain to large instrumental-variable estimates in Switzerland, and the designs estimate effects for different groups of students. Third, employment entry and hiring chances are the most consistently affected outcomes. Fourth, the features most often linked with good outcomes are design features a university or employer can control: a mentor or supervisor with time for the intern, a defined learning purpose, sufficient duration, conditions comparable to regular staff and some formal recognition of what was learned.
What remains uncertain is equally important. No study located here randomizes the design features themselves, so the evidence that mentoring or structure causes better outcomes is associational. Most causal studies come from Europe, and few follow graduates beyond five years. Institutional claims that "our internships lead to jobs" should therefore be read as descriptions of association unless the institution shows a credible comparison group.
An internship design and evidence checklist
The checklist below is an editorial framework proposed by this article for reviewing an internship program; it is not the UNIRANKS methodology. It separates what a program provides from what it can show.
| Area | Question to ask | Stronger evidence | Weaker evidence |
|---|---|---|---|
| Learning purpose | Is there a written learning plan agreed by student, supervisor and academic staff? | Plans on file and reviewed at the end of the internship | A general statement that internships "build skills" |
| Supervision | Does each intern have a named supervisor or mentor with time allocated? | Supervisor guidance or training, and intern feedback on supervision | Supervision assumed but not recorded |
| Duration and conditions | Is the internship long enough to do substantive work, with conditions similar to regular staff? | Recorded duration, hours, pay and insurance by placement | Duration unknown or highly variable |
| Academic integration | Is the experience connected to assessment or structured reflection? | Credit-bearing reflection or assessment linked to learning outcomes | No link between the internship and the program |
| Outcome measurement | Are outcomes measured for participants and comparable non-participants? | Comparison groups, prior attainment controls, or use of changes in requirements | Outcomes reported for interns only |
| Coverage | What share of students participates, and who does not? | Participation reported by program and cohort | Counts of placements without a denominator |
Practical steps
For university administrators and program designers
- Treat the internship as a learning design, not only a placement: define its purpose, the tasks that support it and how learning will be recognized.
- Agree minimum supervision expectations with host organizations and give supervisors short guidance on feedback and task scaffolding.
- Record duration, pay, supervision and academic integration for each placement so that these features can later be linked to outcomes.
- Report outcomes for non-participants alongside participants, and state plainly that differences are associations unless selection has been addressed.
For researchers and evaluators
- Use natural variation, such as a program introducing or removing an internship requirement, or staggered rollouts across departments, to build credible comparison groups.
- Where programs are oversubscribed, consider randomized allocation of places or of design features such as mentoring.
- Measure prior attainment and career intentions before the internship, so that later outcome gaps can be adjusted for them.
- Follow graduates beyond the first year; most current evidence ends at five years or earlier.
For students and families reading program information
- Ask how internships are supervised and assessed, not only how many are offered.
- Treat outcome figures that compare interns with non-interns as descriptive unless the institution explains how it accounted for differences between the groups.
Limitations
This article draws on ten peer-reviewed studies and two official reports. Three of the studies (Bolli et al., 2021; McKenzie et al., 2016; Jones et al., 2017) and both reports were read in full; the other seven studies were read from their abstracts only, so their methods and results are described no further than their abstracts allow. The causal studies come mainly from Germany, Switzerland, Belgium, Austria, Spain and Yemen, and their results may not transfer to other labor markets or to internships organized differently. Two résumé audit studies often cited in this literature are mentioned only through another paper's summary because their publisher pages could not be verified. Questions of who can access internships are outside the scope of this article.
Related reading on UNIRANKS
For how graduate outcomes are defined and measured, see How Should Universities Measure Graduate Employability? and Graduate Employment Rates: Why the Numbers Need Context. More research on this topic is collected on the Employability & Career Success hub.
Editorial disclosure
UNIRANKS publishes university rankings, and its methodology includes an Employability & Career Success dimension. This article reviews independent research on internships and graduate career outcomes. It does not describe how UNIRANKS scores institutions.
Research References
- Bittmann, F., & Zorn, V. S. (2020). When choice excels obligation: about the effects of mandatory and voluntary internships on labour market outcomes for university graduates. Higher Education, 80(1), 75-93. https://doi.org/10.1007/s10734-019-00466-5 (verified 22 November 2026)
- Bolli, T., Caves, K., & Oswald-Egg, M. E. (2021). Valuable Experience: How University Internships Affect Graduates' Income. Research in Higher Education, 62(8), 1198-1247. https://doi.org/10.1007/s11162-021-09637-9 (verified 22 November 2026)
- Di Meglio, G., Barge-Gil, A., Camiña, E., et al. (2022). Knocking on employment's door: internships and job attainment. Higher Education, 83(1), 137-161. https://doi.org/10.1007/s10734-020-00643-x (verified 22 November 2026)
- Hora, M. T., Chen, Z., Wolfgram, M., et al. (2024). Designing Effective Internships: A Mixed-Methods Exploration of the Sociocultural Aspects of Intern Satisfaction and Development. The Journal of Higher Education, 95(5), 579-606. https://doi.org/10.1080/00221546.2023.2241330 (verified 22 November 2026)
- Jackson, D., & Bridgstock, R. (2021). What actually works to enhance graduate employability? The relative value of curricular, co-curricular, and extra-curricular learning and paid work. Higher Education, 81(4), 723-739. https://doi.org/10.1007/s10734-020-00570-x (verified 22 November 2026)
- Jackson, D., & Dean, B. A. (2023). The contribution of different types of work-integrated learning to graduate employability. Higher Education Research & Development, 42(1), 93-110. https://doi.org/10.1080/07294360.2022.2048638 (verified 22 November 2026)
- Jones, C. M., Green, J. P., & Higson, H. E. (2017). Do work placements improve final year academic performance or do high-calibre students choose to do work placements? Studies in Higher Education, 42(6), 976-992. https://doi.org/10.1080/03075079.2015.1073249 (verified 22 November 2026)
- Margaryan, S., Saniter, N., Schumann, M., et al. (2022). Do Internships Pay Off? The Effects of Student Internships on Earnings. Journal of Human Resources, 57(4), 1242-1275. https://doi.org/10.3368/jhr.57.4.0418-9460R2 (verified 22 November 2026)
- McKenzie, D., Assaf, N., & Cusolito, A. P. (2016). The demand for, and impact of, youth internships: evidence from a randomized experiment in Yemen. IZA Journal of Labor & Development, 5(1), 1. https://doi.org/10.1186/s40175-016-0048-8 (verified 22 November 2026)
- Tobback, I., Verhaest, D., & Baert, S. (2024). Internships, Hiring Outcomes and Underlying Mechanisms: A Stated Preferences Experiment. De Economist, 172(1), 25-48. https://doi.org/10.1007/s10645-023-09432-0 (verified 22 November 2026)
Sources
- O'Higgins, N., & Pinedo, L. Interns and outcomes: Just how effective are internships as a bridge to stable employment? EMPLOYMENT Working Paper No. 241 (2018). International Labour Organization. https://www.ilo.org/publications/interns-and-outcomes-just-how-effective-are-internships-bridge-stable (accessed 22 November 2026)
- Atfield, G., Hunt, W., & Luchinskaya, D. Employability programmes and work placements in UK higher education. Research report (November 2021). Department for Education. https://www.gov.uk/government/publications/employability-programmes-and-work-placements-in-higher-education (accessed 22 November 2026)
