Bangladesh Factory Collapse
A garment factory on the outskirts of Dhaka, Bangledesh collapsed on Wednesday, April 24 2013, killing more than 400 and injuring hundreds more. Excerpts from an AP news blog provide some timing information and a glimpse into the horrible tragedy.
Around 8:40 a.m. Wednesday, when the factories had been running for 40 minutes or so, the lights suddenly went off in the building. It was nothing unusual. Bangladesh's electricity network is poorly maintained and desperately overburdened. Rana Plaza, like most of the factories in the area, had its own backup generator, sometimes used dozens of times in a single day.
A jolt went through the building when the generator kicked on. Again, this was nothing unusual. Eighteen-year-old Baezid was chatting with a friend as they checked an order of short-sleeved shirts.
He'd come from the countryside with his family - mother, father and two uncles - just seven months earlier. Since then, he'd worked seven days a week, from 8 a.m. to midnight. His salary was about $55 a month. But he could more than double that by working so many hours, since overtime pays .37 cents an hour.
Sometime after the generator switched on - perhaps a few moments later, perhaps a few minutes - another, far larger, jolt shook the floor violently. The building gave a deafening groan.
The pillars fell first, and one slammed against Baezid's back. He was knocked to the floor, and found himself pinned from the waist down, unable to move.
The GCP event was set for 6 hours starting about when the collapse occurred, which was apparently around 9:00 local time. The result is Chisquare 21516 on 21600 df, for p = 0.657 and Z = -0.404.
I was quite uncertain about the time of the collapse, and was tempted to set the full day of the 24th. But instead I used the standard 6 hour protocol, beginning at my best guess about the time of the collapse. For context, the following exploration shows the 24 hour UTC day. The collapse occurred about 03:00 UTC.
It is important to keep in mind that we have only a tiny statistical effect, so that it is always hard to distinguish signal from noise. This means that every "success" might be largely driven by chance, and every "null" might include a real signal overwhelmed by noise. In the long run, a real effect can be identified only by patiently accumulating replications of similar analyses.