Optimising Cross-Institutional Comparisons in Large-Scale Student Surveys: An Illustration with the National Student Survey (NSS) in Russell Group Institutions

Student surveys have become a primary tool for assessing quality in Higher Education. However, methods used to score and compare survey scores across groups (e.g., institutions) are often suboptimal and based on untested assumptions. While these limitations apply to various student experience surveys, we focus on the National Student Survey (NSS) as an illustrative example. Given the widespread use of NSS data for ranking and comparing institutions—especially the highly competitive group of 24 Russell Group universities in the UK—it is essential to establish the optimal factorial structure of the NSS and examine whether meaningful cross-institutional comparisons can be made. To this end, we used raw, individual level, 2024 NSS data from 95,421 first-degree students at Russell Group institutions to test factorial structure and approximate measurement invariance. Using both CFA and ESEM techniques, we found the NSS is better represented by six rather than the originally proposed seven theme measures (and 22 rather than 24 core questions). Using a contemporary multi-group alignment methodology, we also found this more parsimonious version of the NSS has minimal non-invariance. With evidence of approximate invariance, we provided a revised NSS ranking system based on alignment optimisation. This approach to ranking is more optimal than traditional methods as it uses factor means rather than raw index variables and includes information to cluster institutions with similar factor means. Our results have important implications for how we score and compare scores on the NSS—and other large-scale student experience and engagement surveys—in the future.