Student mental health was a public health problem long before the pandemic. However, first one needs to define what mental health means. Fried (2017) showed that there was a great heterogeneity in the symptoms assessed by the depression scales. Similarly, other factors (study design, assessment period, chosen scale and chosen cut-off or threshold, response rate, data management, etc.) can have an impact on the prevalence found. In addition, theoretical and modelling considerations on mental health need to be answered.

The french Observatoire de la Vie Etudiante measured an increase in student psychological distress since 2016, particularly between 2020 and 2021. Data from three surveys conducted in 2016 (n=18,875), 2020 (n=60,014), and 2021 (n=4,901, longitudinal follow-up from 2020) were used to model psychological distress as a latent common cause or a network with emergent properties.

Preliminary results show that from a latent perspective, measurement invariance does not hold. From a network perspective, the modelled systems showed differences in three aspects (van Borkulo et al., 2022). For participants in the 2020 and 2021 surveys, an increased vulnerability of the modelled system was observed. Prevention and intervention targets in the system were tested with simulation techniques (Lunansky et al., 2022).

Caution is advised for prevalence comparisons when measurement invariance does not hold. The network approach offers an alternative to studying psychological distress as an emergent property of a complex system. However, regardless of the statistical approach, with subjective measures and without measurement error control and qualitative data or cognitive interviews: it is difficult to partition between a change or increase in the phenomenon we wish to measure and a change in the way people tend to respond to a questionnaire, since the representation they have of the specific questions designed to describe this phenomenon might also have changed.

None Declared