tim's experiment

Our beloved Post was in a right tizz this morning over the number of accidents involving minibuses. “Minibuses’ high crash rate fails to spur action,” complained the lead headline. The story started with the news that the accident rate for minibuses was 7.5 times that of all other Hong Kong vehicles, said that this “translated” into 21 deaths and 187 injured last year, and proceeded to accuse the Transport Department of dragging its feet despite the “alarming figures”.

Further statistics floated in further down the story. We were told that the accident rate for minibuses was 255.2 per thousand vehicles, compared with 34.1 per thousand for all vehicles. Alert readers would have spotted at this point that they had already been misled. The minibus accident rate is not 7.5 times the rate for all other vehicles, it is 7.5 times the average for all vehicles including minibuses.
Looking at it another way we were told that only 0.76 per cent of the vehicles in Hong Kong are minibuses, but they are involved in 5 per cent of all accidents. And the question which all this left in my mind is where are the alarming figures? We have all heard the rumours that, as the story put it in due course, “minibuses are notorious for speeding and violating traffic regulations”. I have heard, indeed, that the late night express to Sheung Shui is the most exciting ride in Hong Kong outside Ocean Park. But if we are going to hang a story on the statistics they had better be the right ones. And these weren’t.

Look at it this way. The number of accidents per minibus is bound to be higher than the average. Most Hong Kong cars do a half-hour drive to the office in the morning, and half an hour back in the evening. Much of the “driving” actually consists of sitting in tunnel queues, where a serious accident is scarcely possible because everyone is moving so slowly. Some car drivers don’t even do that. The owner commutes by public transport and drives for fun at the weekend. Minibuses are on the road all the time. Of course the number of accidents per minibus will be higher. I expect we would find that the rate for taxis and buses is also above average, for the same reason. If we want to know whether minibuses are unreasonably dangerous then we need another figure. We could try the number of accidents per mile travelled. Or we could use that great favourite of the airlines the number of accidents per passenger mile. Or if we prefer we could try the number of accidents per trip. I am quite willing to believe that whichever figure you used the outcome would be alarming. But if you are going to beat a government department over the head with the statistics you need a figure which means what you say it means.