What you see may not be what you get

Wednesday, 13 July 2011

Human beings are perhaps never more frightening than when they are convinced beyond doubt that they are right, Laurens van der Post once wrote. I’ve seen this proven many a time, most recently with a close family member (whom I should probably not name, so I won’t tell you that I’m talking about my brother).

I was down in Durban for the weekend. We had a great game of thunee where my sister-in-law and I trounced my brother and another family member (whom I should also probably not name, so I won’t tell you that I’m talking about my cousin).

The conversation over dinner thereafter swung around to youngsters of today and their driving habits. My brother proceeded to recount in vivid detail a few fatal crashes in his recent memory.

The problem, he said, was young Indian boys who get into well-paying jobs, buy top-of-the-line performance vehicles without knowing how to control those vehicles, and proceed to wipe themselves out. And this, he insisted, was a growing phenomenon.

“You’re wrong,” I told him. “This is in the same category as attorneys who get struck off the roll. How many of those lawyers have you heard of that are not Indian?”

His answer, of course, was none.

“That’s because most of the reporting you see on lawyers being struck off the roll comes from Indian reporters who naturally report only on people that they know about, so you almost never hear of the many more attorneys from other ethnic groups who get struck off the roll. So you assume Indian lawyers are more crooked than others,” I said.

“But you don’t live here, “ he said. “I do, and I see what’s happening.”

Observational bias. It leads to a particular problem of not being able to see the forest because there are these trees blocking your view. So let’s get our heads around exactly what happens in the road accident space:

Firstly, KwaZulu Natal has consistently recorded the greatest decrease in number of fatal crashes around the country. Year on year deaths have declined by more that 500 per year which translates to about 22 percent. Now consider that the number of cars on KZN roads increased by about 5 percent in the same period. So the number of people getting killed has decreased.

Secondly, pedestrians are the single biggest factor in road deaths. They account for about 44 percent of those killed on our roads nationally and for more than 50 percent of those killed in KZN. Thereafter, the causes of fatalities are overtaking manoeuvres (26 percent), failure to stop or yield (7 percent), and unsafe turning (5 percent). Nowhere in that lot do you see “boy racer losing control of new powerful vehicle”.

Thirdly, the biggest number of fatalities per age group is among 30-34 year olds, and this applies to drivers, passengers, and pedestrians alike. Deaths in the 25-29 age group are roughly the same as the 35-39 age group, and deaths in the 20-24 age group are few than in the 40-44 age group. Most importantly, the number of deaths in those younger groups are decreasing at a faster rate than among the older generation.

So we need to be wary of drawing inferences based on what we can see, because observational bias very quickly leads to confirmation bias. For example, you get a new car, and suddenly start noticing whenever you see the same model of car.

So if you expect to hear of Indian lawyers being struck off the roll or of young Indian professionals wiping themselves out in their new performance vehicles, that is what you will see. But don’t take that as a representative sample.

Finally, the last type of observational bias to be wary of is known as the “streetlight effect”. The story goes that a policeman sees a drunkard searching for something under a streetlight. “I’ve lost my car keys,” says the drunkard. The policeman then begins to look around as well. “Are you sure you lost them here?” he asks. “No”, says the drunkard, “I lost them in the park.” “So why the hell are we looking here?” asks the policeman. “Because this is where the light is,” says the drunkard.

So you might chuckle at the idiocy at the drunk, but then you’ve missed the stupidity of the policeman who should have asked why a drunkard needed car keys.