Method & mapping

How the WEC is read through the PEM framework, and the choices behind it.

The framework

This overview interprets the data through the PEM model of student well-being (Murray et al., 2024), the same model that anchors the Equilibria approach. Well-being is treated not as a score to be maximised but as a dynamic balance across three internal dimensions:

  • Physical — the capacity to maintain a healthy quality of life and daily functioning.
  • Environmental — the value a student places on the links between themselves, their spaces and their people.
  • Mental / Emotional — the capacity to cope with the stresses of life.

These dimensions are acted on by three external influences: relationships (the central mediator), perivallon (the physical surroundings), and student voice. The influences relate to the dimensions through reinforcing feedback loops rather than simple one-way causes.

The data

The source is the Wellbeing & Engagement Collection (WEC), South Australia’s annual Years 4–12 student survey (Gregory & Brinkman, 2014). Each subdomain summarises several Likert items into the share of students reporting high, medium or low well-being. The WEC is a population-level snapshot; it does not track individuals over time.

This overview is built from the year-level state collections for 2020–2024. A few points to keep in mind:

  • Years 4–12 only. Totals therefore run ~1–2% below the official “All SA” figures, which also include “Primary/Secondary Other” and “Unknown” groups. No top-up has been applied.
  • 2020 was a Term-3, COVID-affected collection and should be read with that caveat.
  • 2018–2019 data exist in the source and sit several points higher on most dimensions; they are held back from the headline window but are available for an extended-baseline appendix.
  • Gender is not included in this first version; gender splits exist only for a subset of subdomains and are deferred to a later iteration.

Mapping the WEC to PEM (Table 1)

Each WEC subdomain is assigned to one PEM construct. The mapping follows Murray et al. (2024); the placements marked with a dagger (†) are defensible but worth a conscious sign-off, and are flagged here for transparency.

PEM role Construct WEC subdomains
Dimension Physical Sleep · Sports · Nutrition (breakfast)
Dimension Environmental Connectedness to school · Peer belonging · School climate · School belonging
Dimension Mental / Emotional Happiness · Optimism · Sadness · Worries · Emotion regulation
Factor Relationships Academic self-concept † · Cognitive engagement † · Perseverance † · Social/Verbal/Physical/Cyber bullying · Friendship intimacy · Emotional engagement with teachers
Factor Perivallon Music and arts · Organised activities
Factor Voice (no WEC indicator)
— Unmapped Satisfaction with life

Four choices worth noting:

  1. Learning-readiness items under Relationships (†). Academic self-concept, cognitive engagement and perseverance are placed with the relationships factor. They read at least as much like learning readiness; their placement is the most contestable in the table.
  2. All four bullying types under Relationships. Defensible (bullying is relational), but it means the relationships factor mixes supportive and adverse relational experiences; the subdomain page keeps them separate so the two do not cancel out.
  3. Satisfaction with life is left unmapped. It does not sit cleanly inside any one PEM construct and is excluded from the aggregates rather than forced.
  4. Voice has no indicator at all. The WEC simply does not measure student voice. We report this as a finding rather than papering over it.

Research questions

  1. RQ1 — change over time. How has the proportion of students reporting high well-being changed across 2020–2024, by subdomain and by PEM construct?
  2. RQ2 — the developmental gradient. How does well-being differ across year levels (Years 4–12) in 2024?

Trend classification

Each subdomain’s statewide time series is classified into a trend type, following the school-report scheme:

  • Type 0 — broadly stable (range under ~2 percentage points).
  • Type 1+ / 1− — steady improvement / steady decline (linear fit adequate).
  • Type 2 — quadratic: a dip-then-recovery or rise-then-fall.
  • Type 3 — cubic, with multiple turning points.

With only five annual points in the 2020–2024 window, classification is capped at Type 2: a cubic fit on five points is effectively saturated and would overstate structure. Type 3 detection is reserved for the longer 2018–2024 series, which can be enabled as an extended-baseline view.

Reproducibility

The whole overview is generated from a single tidy dataset (wec_state_tidy.csv: year × year level × subdomain × high/medium/low, tagged with PEM role and construct), so every figure and table here, and the state context used in each school report, derive from the same source. Re-running the build script after a new WEC release refreshes the entire site.

References

Gregory, T. & Brinkman, S. (2014). Wellbeing and Engagement Collection. Government of South Australia.

Murray, C., Gabriel, F. & Kennedy, J. (2024). Factors that promote student well-being in schools: a scoping review of Australian and Aotearoa New Zealand literature. Humanities and Social Sciences Communications, 11:1542. https://doi.org/10.1057/s41599-024-04059-1