Universities can measure multicultural inclusion only by looking past who is enrolled to how students experience the institution. Useful measures cover belonging, the amount and quality of interaction between domestic and international students, experiences of discrimination and what happens after they are reported, access to teaching and services in a language students can use, participation in class, activities and governance, and whether the curriculum is internationalized for all students. Survey-based measures should use appropriately validated instruments, report response rates and be examined across relevant student groups. Administrative and institutional measures require clear definitions, reliable records, privacy safeguards and transparent reporting. A nationality distribution shows diversity. It does not show inclusion.
Diversity is a count; inclusion is an experience
Demographic diversity describes the composition of a student body: how many students hold each citizenship, come from each region or speak each first language. An inclusive environment describes conditions: whether students from different backgrounds feel they belong, interact with each other, take part in learning and student life, are protected from harassment, and have problems resolved fairly. The two can move independently. A campus can enroll students from many countries while those students study, socialize and seek help in separate groups.
Spencer-Oatey and Dauber (2019) make this point directly. Their article argues that internationalization is often assessed through compositional criteria, such as the numbers of international students and staff, and that other measures are needed if diversity is to improve the quality of education. Enrollment diversity itself, including concentration and country of origin, is covered in our article on why international student enrollment numbers are not enough. This article concentrates on the inclusion side of the equation.
What the research shows
Importance and experience of intercultural mixing are different things
Spencer-Oatey and Dauber (2019) surveyed 2,360 students in four countries (1,455 domestic, 265 from the European Economic Area and 640 from Asia, in the authors' classification) using a questionnaire that asks students to rate each item twice on 6-point scales: how important it is to them and how much they actually experience it. The constructs were social integration, academic integration, and global opportunities and support. For each construct, ratings of importance and experience differed significantly, and the scatterplots showed wide variation both within and across regional groups. The authors note that overall means mask this variation and conclude that a single approach for all students will be inadequate. They also state that their two scales differ conceptually from satisfaction ratings and that combining them gives a more differentiated view of where satisfaction comes from.
Belonging differs among international students, not only between international and domestic students
Mullen and Li (2025), using survey data from a large university in southern Ontario, compared three groups of international students (Chinese, South Asian and students from other countries) with domestic students. According to the abstract, Chinese international students reported a greater sense of belonging than domestic students, South Asian students did not differ, and students from other regions reported a more tenuous sense of belonging. The authors link part of these differences to background characteristics, involvement in campus activities and cross-racial socialization, and report that international students' disproportionate experiences of on-campus racism and discrimination were associated with lower belonging in their analysis. For measurement, the lesson is that a single "international students" category can hide groups whose experiences point in opposite directions.
Positive climate averages can coexist with frequent discrimination
Hanft-Robert and colleagues (2026) surveyed 987 students in health-related degree programs at 83 German universities across all 16 federal states between January and April 2024. Mean ratings of the diversity climate were moderately to highly positive: 3.91 of 5 for age, 3.63 for ethnicity and 3.83 for gender. Yet 28% of respondents reported personally experiencing discrimination at their university in the previous two years, and 45% reported witnessing it. Across the sample, 61% said teaching materials did not adequately represent diverse groups, and 50% perceived lecturers as insufficiently responsive to discrimination. Students who had experienced discrimination reported a less positive climate and a lower sense of belonging, among other differences. The authors recruited through convenience and snowball sampling and state that the sample is not representative, so the percentages describe this sample only.
Discrimination is associated with weaker social contact
Balcı, Bozdağ and Serin (2026) studied 448 international students in Türkiye, comparing students from the Turkic Republics with other international students. In their structural equation models, perceived discrimination was associated with poorer social adjustment through lower quantity and quality of social contact, and the role of contact quality held across both groups. The authors state that their cross-sectional design precludes causal inference. Their findings support measuring contact quality, not only contact frequency, and reading discrimination and interaction data together rather than in separate reports.
Structured contact programs: modest and mixed evidence
Dong, Bryant and Liu (2024) evaluated a Cultural Partner Program at a public research university in the southeastern United States. Of 97 domestic graduate students, 68 were assigned to the experimental group and paired with an incoming international graduate student for one semester; the remaining 29 formed a control group. Regression analyses found a significant conditional difference in wellbeing but not in the other outcomes, while standardized mean differences suggested improvements in perceived support, interaction attentiveness and stress. A single-site study of this size cannot establish general effects, but it shows how an institution can evaluate an intercultural initiative with pre- and post-tests and a comparison group instead of counting participants.
Curriculum, teaching and the wider climate
Beelen and Jones (2015) propose a working definition of internationalization at home as "the purposeful integration of international and intercultural dimensions into the formal and informal curriculum for all students" within domestic learning environments. They stress that electives alone do not reach all students and that internationalization at home does not require the presence of international students. This shifts measurement from who is present to what every student is taught and assessed on. Hurtado and colleagues (2012), in a chapter read at abstract level, present a model in which many aspects of an institution influence and contribute to the campus climate for diversity, which in turn plays a role in student outcomes. Both works treat inclusion as an institutional condition rather than a property of individual students.
Representation and voice are not the same as partnership
Matthews and Dollinger (2023) distinguish elected student representation from staff-selected student partnership. They warn of harm in positioning student partners as speaking for other students and of the risk of weakening elected representation systems. For inclusion measurement, this means that counting international or minority students on committees is not sufficient; evaluators also need to know how those students were chosen, whom they represent and whether their input changes decisions.
Why satisfaction surveys and demographic percentages mislead
Non-response
Standish and Umbach (2019) linked college student survey responses to administrative records on facility use, academic performance, class attendance and co-curricular participation. Response propensity was related to actual behavior for 2 of 3 survey topics, and respondents and nonrespondents differed significantly on 11 of 13 behavioral measures. The authors conclude that survey summary statistics may not generalize to the target population. Applied to inclusion, the students least connected to campus life may be the least likely to answer an inclusion survey, which would make results look better than the underlying reality.
Response bias and ceiling effects
Satisfaction questions can also suffer from problems that response rates cannot fix. Students may give socially acceptable answers, may hesitate to criticize the institution that teaches and assesses them, and may answer differently depending on who administers the survey. When most respondents choose the top categories, a ceiling effect leaves little room to detect improvement or deterioration, and a small group with very negative experiences barely moves the mean. The German study above illustrates the general pattern: a positive average climate rating sat alongside a 28% discrimination rate in the same sample.
Measurement invariance across cultural and language groups
Comparing scores between groups assumes that the questions measure the same thing in the same way for each group. Zavala-Rojas and Saris (2018) tested this for items on political trust and satisfaction with politics across six linguistic groups in four countries, using European Social Survey data. They report that concepts were only partially invariant, and only once differences in how groups use the response scale were allowed for. Their subject is not higher education, but the implication carries over: if an inclusion survey is offered in several languages, or taken by students with different response styles, differences in mean scores may reflect measurement rather than experience unless invariance has been tested.
What demographic percentages cannot show
A share of international students, or a count of nationalities, cannot show whether students interact, whether they are treated fairly, whether they can use services, or whether they stay and complete. Percentages also depend on definitions (citizenship, prior residence or fee status) that differ across systems. They remain useful as context and as a denominator for other measures, such as the share of each group that reports discrimination or holds an elected role.
Reporting, response and complaint resolution: what public bodies require
Some regulators now specify how institutions must handle harassment, which gives evaluators concrete features to check. The Office for Students (OfS), which describes itself as the regulator for higher education in England, applies condition E6 on harassment and sexual misconduct to all registered providers; according to the OfS, it came into force on 1 August 2025. Providers must maintain a single comprehensive source of information on their policies and procedures, which must set out the ways in which students, staff and others can report behavior, the support provided to students, how investigations and decisions are made credible and fair, and how staff handling cases are trained. Condition E6.10 prohibits restricting provisions in contracts formed or varied after the condition takes effect. The condition defines harassment by reference to section 26 of the Equality Act 2010, whose list of relevant protected characteristics includes race and religion or belief.
At the European level, the principles and guidelines on the social dimension annexed to the Rome Ministerial Communiqué of the European Higher Education Area call for national data on the composition of the student body, access, participation, drop-out and completion, complemented where needed by institutional surveys. They also ask public authorities to consider ombudsperson-type institutions able to mediate conflicts, particularly those related to equity, and suggest that external quality assurance address how diversity and inclusion are reflected in institutional missions. Neither document sets an inclusion score, but both show that reporting channels, support, fair procedures and data are expected institutional features, not optional extras.
For evaluators, the key distinction is between the existence of a reporting route and evidence that it works: how many reports were received, how long resolution took, what proportion of complainants were offered support, and whether reporters from different groups trust the process. A low number of reports can mean few incidents or low confidence in reporting; it cannot be interpreted without survey data on experiences.
An editorial framework for measuring multicultural inclusion
The table below is an editorial framework proposed by this article. It is not the UNIRANKS methodology and does not describe how UNIRANKS evaluates institutions.
| Measure | What it can show | What it cannot show | Minimum quality condition |
|---|---|---|---|
| Nationality or citizenship distribution | Composition and concentration of the student body | Interaction, fairness, belonging or outcomes | Stated definition (citizenship, residence or fee status) and reference date |
| Sense of belonging scale | Whether students feel accepted and connected, by group | Why belonging is low, or which practice caused a change | Validated instrument, disaggregation beyond "international vs domestic", response rate reported |
| Intercultural interaction (importance and experience) | Gaps between what students want and what they experience | Quality of learning from contact on its own | Separate importance and experience items; results shown as distributions, not only means |
| Contact quality and frequency | Depth of relationships across groups | Direction of cause between contact and adjustment | Both quantity and quality measured; repeated over time |
| Experiences and witnessing of discrimination | Prevalence of harmful experiences by group | Whether incidents were reported or resolved | Defined reference period, confidential administration, group-level results with small-cell suppression |
| Reports and complaint outcomes | Use and handling of formal procedures | True incidence (low reporting can mean low trust) | Read alongside survey prevalence; resolution times and support offered recorded |
| Language accessibility | Whether policies, reporting routes and key services are usable by students who are not fluent in the main language | Whether students actually understood or used them | Inventory of languages offered plus user feedback; surveys tested for invariance across language versions |
| Academic participation and inclusive teaching | Participation in class and group work; representation in teaching materials; staff response to incidents | Learning gains on its own | Student-reported and course-level evidence; curriculum review covering compulsory courses, not only electives |
| Representation in activities and governance | Whether different groups hold roles and have a voice | Whether that voice influences decisions | Distinguish elected representation from appointed partnership; record how roles were filled |
| Satisfaction ratings | General sentiment among respondents | Experiences of non-respondents; differences hidden by ceiling effects | Response rate and respondent profile published; top-box and bottom-box shares shown with means |
Our interpretation
Read together, these studies suggest three conclusions. First, averages are the weakest form of inclusion evidence. The German climate study, the Canadian belonging study and the four-country integration study all found that aggregate figures concealed groups with markedly different experiences. Second, inclusion evidence is most informative when measures are linked: discrimination experiences, contact quality and belonging were related to each other in the studies that measured them together, although none of these designs can establish cause. Third, institutional processes, such as reporting routes, fair investigations, curriculum design and elected representation, are what an institution directly controls, and they can be documented independently of survey sentiment.
A reasonable evaluator would therefore give more weight to a university that publishes disaggregated belonging and discrimination results with response rates, and shows how it acted on them, than to one that publishes a high satisfaction score or a long list of nationalities. That is a judgment about evidence quality, not a claim about any ranking outcome.
Practical steps for universities and evaluators
- Separate composition from inclusion. Report demographic data as context and use them as denominators, but never present them as evidence of inclusion.
- Use validated instruments and test them. Choose belonging and climate scales with published validation evidence, and test measurement invariance across language versions and major student groups before comparing scores.
- Disaggregate carefully. Break results down beyond "international" and "domestic", while suppressing small groups to protect privacy.
- Publish the survey's quality, not only its results. State response rates, the respondent profile, the reference period and the share of answers at the top and bottom of each scale.
- Pair experience surveys with process data. Read discrimination prevalence alongside formal reports, resolution times and support offered, so that low reporting is not mistaken for low incidence.
- Check language accessibility. Make policies, reporting routes and key services available in the languages students use, and ask students whether they could use them.
- Audit the curriculum and teaching. Review compulsory courses for international and intercultural content, and ask students whether teaching materials and staff responses are inclusive.
- Evaluate initiatives, do not just count them. Use pre- and post-measures and comparison groups for buddy, partner or intercultural programs, as Dong and colleagues did.
- Record how student voice works. Document how representatives are elected or selected and what changed as a result of their input.
Limitations
This review draws on ten research works. Six were read in full (Spencer-Oatey and Dauber; Hanft-Robert and colleagues; Balcı and colleagues; Dong and colleagues; Matthews and Dollinger; Beelen and Jones) and four from the abstract and publisher page only (Mullen and Li; Standish and Umbach; Zavala-Rojas and Saris; Hurtado and colleagues). Two of the ten are book chapters rather than journal articles. The empirical studies come from single universities or single national systems (Canada, Germany, Türkiye, the United States and four countries in the integration study), several use convenience samples, and all quantitative designs are cross-sectional or small-scale, so none supports causal claims about what improves inclusion. The measurement invariance study concerns general social survey items, not higher education instruments. Regulatory examples are limited to England and the European Higher Education Area and are described, not interpreted as legal advice. Peer-reviewed evidence on how institutions handle discrimination complaints, as distinct from prevalence, remains thin in the sources we could verify.
Related reading on UNIRANKS
For the full set of articles in this series, see the Internationalization & Diversity research hub. Support services designed specifically for international students, which overlap with complaint handling and language help, are examined in our companion article on what support universities should provide international students. Because belonging and wellbeing are closely related, our guide to measuring student wellbeing and facilities covers the wellbeing side of the same evidence.
Editorial disclosure
UNIRANKS publishes university rankings, and its methodology includes an Internationalization & Diversity dimension. This article reviews independent research on measuring multicultural inclusion in universities. It does not describe how UNIRANKS scores institutions.
Research References
- Balcı, H. A., Bozdağ, F., & Serin, H. (2026). Beyond cultural distance: perceived discrimination, social contact, and social adjustment among international students in Türkiye. BMC Psychology, 14(1), 825. https://doi.org/10.1186/s40359-026-04585-4 (verified 21 November 2026)
- Beelen, J., & Jones, E. (2015). Redefining Internationalization at Home. In The European Higher Education Area (pp. 59-72). Springer. https://doi.org/10.1007/978-3-319-20877-0_5 (verified 21 November 2026)
- Dong, S., Bryant, C., & Liu, L. (2024). Fostering intercultural interactions and outcomes for domestic graduate students through internationalization-at-home efforts. Social Psychology of Education, 27(4), 2059-2078. https://doi.org/10.1007/s11218-024-09902-6 (verified 21 November 2026)
- Hanft-Robert, S., Adedeji, A., Junger, A., et al. (2026). Diversity climate and discrimination at German universities: a cross-sectional study among students in health-related degree programs. BMC Medical Education, 26(1), 1095. https://doi.org/10.1186/s12909-026-09806-3 (verified 21 November 2026)
- Hurtado, S., Alvarez, C. L., Guillermo-Wann, C., et al. (2012). A Model for Diverse Learning Environments. In Higher Education: Handbook of Theory and Research (pp. 41-122). Springer. https://doi.org/10.1007/978-94-007-2950-6_2 (verified 21 November 2026)
- Matthews, K. E., & Dollinger, M. (2023). Student voice in higher education: the importance of distinguishing student representation and student partnership. Higher Education, 85(3), 555-570. https://doi.org/10.1007/s10734-022-00851-7 (verified 21 November 2026)
- Mullen, A. L., & Li, Y. (2025). Belonging beyond borders: a comparative analysis of international and domestic students at a Canadian university. Higher Education, 90(3), 801-818. https://doi.org/10.1007/s10734-024-01351-6 (verified 21 November 2026)
- Spencer-Oatey, H., & Dauber, D. (2019). Internationalisation and student diversity: how far are the opportunity benefits being perceived and exploited? Higher Education, 78(6), 1035-1058. https://doi.org/10.1007/s10734-019-00386-4 (verified 21 November 2026)
- Standish, T., & Umbach, P. D. (2019). Should We Be Concerned About Nonresponse Bias in College Student Surveys? Evidence of Bias from a Validation Study. Research in Higher Education, 60(3), 338-357. https://doi.org/10.1007/s11162-018-9530-2 (verified 21 November 2026)
- Zavala-Rojas, D., & Saris, W. E. (2018). Measurement Invariance in Multilingual Survey Research: The Role of the Language of the Questionnaire. Social Indicators Research, 140(2), 485-510. https://doi.org/10.1007/s11205-017-1787-x (verified 21 November 2026)
Sources
- Condition E6: Harassment and sexual misconduct. Office for Students. https://www.officeforstudents.org.uk/for-providers/student-protection-and-support/harassment-and-sexual-misconduct/condition-e6-harassment-and-sexual-misconduct/ (accessed 21 November 2026)
- Who we are. Office for Students. https://www.officeforstudents.org.uk/about/who-we-are/ (accessed 21 November 2026)
- Equality Act 2010, section 26: Harassment. legislation.gov.uk (The National Archives). https://www.legislation.gov.uk/ukpga/2010/15/section/26 (accessed 21 November 2026)
- Rome Ministerial Communiqué, Annex II: Principles and Guidelines to Strengthen the Social Dimension of Higher Education in the EHEA. European Higher Education Area. https://ehea.info/Upload/Rome_Ministerial_Communique_Annex_II.pdf (accessed 21 November 2026)
