This company has 1,000 people and a true turnover of 10% a year that never changes. Each panel measures the same replayed years. Only how often you look is different, and a shorter window is simply a smaller sample of the same truth.
Years replayed: 0
Try a fix:
Annual1 reading a year
–of readings outside the reference band
Quarterly, annualized4 readings a year, each quarter × 4
–of readings outside the reference band
Monthly, annualized12 readings a year, each month × 12
–of readings outside the reference band
Rolling 12-month12 readings a year, each covering the last 12 months
–of readings outside the reference band
2 SD reference band: true rate ± 2 SD of the annual number
Readings that would raise an alarm
True rate (never changes)
Expected shape (exact math)
Bar height: how often a reading lands there, scaled to fit each panel. Compare widths, not heights.
Your turn: real change, or just noise?
Part 2. Above, nothing ever changed, and a yardstick built for yearly numbers flagged … of monthly readings in expectation. Here each month is judged the way an XmR chart judges it, and in half the rounds something really does change. Can you tell which?
Read the chart. The grey dots are last year's 12 months. They set the average line and the red limits: average ± 2.66 × the average month-to-month change. The 24 dark dots are this period, each one month's leavers × 12 ÷ headcount. A red dot falls outside the limits.
Make the call. In half the rounds the true rate secretly moves up or down by 25% to 60% in some month between 7 and 18. In the other half it never moves. In stable, evaluable rounds, the simulated share of dark dots turning red is ….
See the truth. Answering reveals the true rate and what the three XmR rules would have flagged. The scoreboard then compares your calls with the rules.
XmR unavailable: all 12 baseline counts are identical, so the moving range is zero. No XmR stability or change conclusion is available; human and oracle calls are still scored.
Observed counts and annualized rates
Fewer than one expected leaver per month: discrete counts make these XmR rules unreliable. A zero-moving-range baseline is not evaluable. Consider methods designed for rare events.
Did the true turnover rate change during these 24 months?
The baseline rate is given above; any future change stays hidden until you answer. Focus this answer area or a choice, then press Y or N.
Benchmark waiting…
Detection timing benchmark
Among evaluable changed rounds, signals before the change are false alarms. A pre-change-only signal counts as “changed” in classification but misses detection. Delay starts at zero in the change month; mean delay includes detected rounds only.
Whole-dashboard classification scoreboard
Benchmark: 2,000 seeded rounds (seed 20260928) at your current settings. Round-level rates show numerator/denominator and Wilson 95% Monte Carlo intervals. These intervals describe simulation uncertainty, not model validity. Answering at random averages 50% in both classification columns; a handful of rounds proves little. Changing a setting starts a fresh scoreboard.
* Approximate oracle: knows the baseline rate and the change model, and weighs all 24 test months together using 64 midpoint quadrature points for the continuous uniform 25–60% shift prior in each direction. It is a model-specific reference, with numerical approximation error. XmR instead estimates its limits from only 12 baseline months.
XmR rules, as listed in Stehlík (2024): (1) a point outside the limits; (2) three of four successive points closer to the same limit than to the average; (3) eight successive points on the same side of the average. Limits here come from the grey baseline year only.