What Does Neutrality Tell Us?
INCIDENT–ROSTER OVERLAP UNDER THE NULL HYPOTHESIS

Chance or Guilt?

Mean / Q95 / Q99 Surfaces
Incidents(M) 40
Snaps to exported grid values
Mean presence (p̄) 25.0%
Continuous interpolation between grid rows
COMPARE OBSERVATION
Observed incidents
Highest overlap count for any nurse
Observed overlaps cannot exceed total incidents
Exceeds chance
This overlap count exceeds what would typically occur under random shift assignment (beyond the 99th percentile). This does not indicate guilt—it indicates that roster-based analysis alone cannot rule out chance, and professional statistical review by a qualified expert (e.g., Royal Statistical Society accredited) is strongly recommended before drawing any conclusions.
Warrants review
This overlap count is elevated but not extreme. It may warrant internal review to understand rostering patterns, but does not by itself indicate wrongdoing. Consider consulting with departmental leadership.
Within expectation
The observed overlap count falls within the range expected under random shift assignment. No statistical anomaly is indicated.
Exact percentile and tail-probability readout will appear when the manifold pack includes the simulated max-overlap distribution.
Loading simulation parameters...
—
Loading manifold data...
Camera preset Isometric
Front-on 3D can visually compress the p̄ slope. Use "Across p" or the slice profile on the right to inspect it directly.
Mouse Explore the manifold
  • Drag anywhere to orbit · Scroll to zoom
  • Drag the X on the floor grid to move the intercept (updates the sliders)
M-SLICE PROFILE Current M across mean presence
Preparing slope summary…
CURRENT INTERCEPT (M=40, p̄=25.0%)
Mean
—
Q95
—
Q99
—
Mean
Q95
Q99
Navigation & Mouse
Drag to orbit · Scroll to zoom · Drag the X to move intercept
Sliders and the top camera dock update the expected max-overlap view.