Every report was scored against each criterion, with a one-line finding, the model's reasoning and the evidence it relied on. The bars show how the reports fall across each scale's bands, lowest left; the grid reads the first two criteria together.
What the model was asked
Assess the severity and systemic reach of the report's adverse findings. Base scores on the findings themselves (deficiencies, failures, consequences), not on the report's tone. A report finding strong assurance with no material issues should score low on both.
Finding severity
Overall severity of the adverse findings (1 = assurance / negligible issues, 3 = significant material deficiencies, 5 = severe failures with major consequences)
3.45mean of 4,183 reports · scale 1–5
Assurance
Material
Severe
By kind of body
Mean
Auditor-General
3.46
Ombudsman
3.46
Anti-corruption / integrity commission
3.6
Royal Commission
4.8
Inspector-General
3.42
Productivity Commission
2.66
Independent review
3.23
Finding severity
Extent to which findings are systemic/pervasive rather than isolated (1 = isolated one-off, 3 = recurring within the entity/program, 5 = structural failure spanning the system or repeated unaddressed over time)
3.88mean of 4,182 reports · scale 1–5
1
2
3
4
5
By kind of body
Mean
Auditor-General
4.06
Ombudsman
3.6
Anti-corruption / integrity commission
3.47
Royal Commission
4.94
Inspector-General
4.01
Productivity Commission
3.4
Independent review
4.12
The first two criteria, together
Each cell counts the reports in both bands. Choose a cell to rank only those reports.
Columns: Overall severity of the adverse findingsRows: Extent to which findings are systemic/pervasive rather than isolated
