Simpson's paradox: the more reliable courier with the worse on-time rate
Two delivery couriers are compared by on-time rate (on-time deliveries / total), split by traffic conditions:
| Off-peak | Rush hour | Overall | |
|---|---|---|---|
| Blue | 95/100 (95%) | 540/900 (60%) | 635/1000 (63.5%) |
| Red | 846/900 (94%) | 55/100 (55%) | 901/1000 (90.1%) |
Blue is more punctual off-peak (95% vs 94%) and in rush hour (60% vs 55%), yet Red's overall on-time rate is far higher (90.1% vs 63.5%).
Which courier is more reliable, and explain the reversal.