Genuine question to the experts: What are the chances that this effect is caused by how we attribute the accidents to time intervals? I haven't read the whole p

Genuine question to the experts: What are the chances that this effect is caused by how we attribute the accidents to time intervals? I haven't read the whole paper, but when they say accident rate is increased by 6%, is that counted by 24-hour intervals? Or by calendar days?