Belonging predictors stayed stable through the pandemic
Updated Jul 13, 2026
Universities often talk about belonging as if it were too fragile to compare across difficult periods. This Higher Education paper suggests something more useful: even through the pandemic, the main predictors of student belonging remained strikingly stable. That matters for teams reviewing belonging survey measures, because it suggests universities should spend less time reinventing the concept and more time acting on the institutional conditions that shape it. Kelly-Ann Allen, Joseph Crawford, Taren Sanders, Bonnie Bozorg and Cassandra Saunders make that case in "Belonging beyond crisis: Predictors are stable in higher education, despite a pandemic", published in Higher Education.
Context and research question
Belonging now sits near the centre of higher education strategy. Universities connect it to continuation, wellbeing, engagement, inclusion, and the broader student experience. But there is still a practical uncertainty behind many institutional dashboards: if the context changes sharply, can teams trust the same belonging signals, or do the drivers shift as well?
Allen and colleagues test that question using a very large Australian dataset. The study draws on the Australian Student Experience Survey, or SES, covering 733,625 students between 2020 and 2022, and extends earlier pre-pandemic work covering 2013 to 2019. Using machine learning, specifically boosted trees modelling, the authors ask whether the pandemic changed the core predictors of belonging or whether the same institutional levers continued to matter.
That question travels well to UK higher education. Institutions here also use belonging as an early warning signal in student success, widening participation, and support planning. If the underlying predictors are stable, the implication is that universities should protect consistent measures and focus on interpreting them well rather than constantly redesigning them.
Key findings
The headline result is that the core predictors of belonging were stable, even during a major disruption. Across the pandemic period, the strongest influences remained overall educational experience, opportunities for student interaction, and support for settling into university life. For UK teams, that matters because these are not abstract traits. They are institutional conditions that can be designed, reviewed, and improved.
The abstract states the point clearly:
"the primary predictors of belonging remained largely consistent with pre-pandemic findings"
The pandemic changed the context, but not the basic logic of belonging. That is a useful corrective for universities tempted to treat crisis periods as analytically exceptional. Students were learning under very different conditions, but the factors most associated with belonging still centred on connection, transition, and the wider educational experience rather than on a wholly new set of variables.
The study also reports a slight increase in model precision during the pandemic period. The authors interpret this as a possible homogenisation of student experience under disruption. In practice, that suggests the pandemic may have compressed parts of university life into a narrower set of common pressures, but it did not displace the importance of peer interaction, institutional support, or a coherent student experience.
The broader significance is methodological as much as substantive. Large-scale survey analysis can identify stable belonging predictors across years, but it still does not explain the mechanism behind a score. A weak belonging result might reflect induction gaps, commuter friction, reduced peer contact, unclear support routes, or something more specific for a subgroup. That is why this paper sits naturally alongside work on survey benchmarking and triangulation: the score shows where to look, but not yet what to fix.
Practical implications
The first implication for UK universities is to keep a stable core belonging framework across years wherever possible. If the main predictors are resilient, institutions gain more from comparability than from repeatedly rewriting their survey logic. A consistent core makes it easier to tell whether a cohort genuinely changed, or whether the institution simply changed the measurement.
Second, universities should prioritise the institutional conditions that repeatedly shape belonging. Settling in, peer interaction, and overall educational experience are all areas where Student Experience teams can intervene through induction design, transition support, timetabling, community-building, and clearer academic support routes. The benefit is practical: teams can focus on levers they can actually move rather than treating belonging as an intangible mood.
Third, institutions should pair belonging scores with open-text evidence that explains the result. A survey can tell you that belonging is weaker for a cohort, but it will not tell you whether the issue sits in social connection, staff approachability, hybrid delivery, or support access. That is where our NSS open-text analysis methodology becomes useful, and where Student Voice Analytics fits naturally. Comment analysis helps universities distinguish between different forms of belonging friction, compare them across cohorts, and avoid overreacting to a single headline measure.
Finally, universities should treat subgroup variation as part of the main analysis, not a later add-on. Stable overall predictors do not mean belonging is experienced evenly. Earlier work on what ethnic-minority students mean by belonging shows why broad averages can miss important differences in how belonging is understood and negotiated. The payoff from a combined survey-and-comment approach is a clearer basis for action that is both trend-aware and sensitive to lived experience.
FAQ
Q: How should a university apply this paper when reviewing its own belonging survey?
A: Keep the core measure stable, then test whether your action planning is focused on the same areas this paper highlights: overall educational experience, peer interaction, and support for settling in. If your survey only produces a headline belonging score, add one or two open-text prompts so teams can see what is driving the result in practice.
Q: What are the methodological limits of this study?
A: The paper uses a very large Australian survey dataset and machine learning to identify predictors, which makes it strong for pattern detection but not a complete substitute for local diagnosis. The results show what is consistently associated with belonging across a large population. They do not by themselves explain how different student groups interpret belonging, or why a particular institutional score changed in one setting.
Q: What does this change about student voice work more broadly?
A: It strengthens the case for treating quantitative and qualitative evidence together. Belonging surveys are useful for monitoring and comparison, but student voice becomes more actionable when institutions can connect a movement in the score to the language students use about friendship, transition, confidence, support, and participation. That is especially important when leaders want evidence that is both consistent over time and specific enough to act on.
References
[Paper Source]: Kelly-Ann Allen, Joseph Crawford, Taren Sanders, Bonnie Bozorg and Cassandra Saunders "Belonging beyond crisis: Predictors are stable in higher education, despite a pandemic" DOI: 10.1007/s10734-025-01549-2
Request a walkthrough
Book a free Student Voice Analytics demo
See all-comment coverage, sector benchmarks, and reporting designed for OfS quality and NSS requirements.
- All-comment coverage with HE-tuned taxonomy and sentiment.
- Versioned outputs with TEF-ready reporting.
- Benchmarks and BI-ready exports for boards and Senate.
UK-hosted · No public LLM APIs · Same-day turnaround