Stephen Dubner: Turkeys and why data is useless without experimentation
Posted on September 27, 2017
6 min read
In a wide-ranging talk at the 2017 Telstra Vantage conference, opening keynote speaker Stephen Dubner — an acclaimed journalist and co-author of the best-selling book Freakonomics — used stories of poultry farming, hand hygiene compliance in hospitals, and even rat plagues to illustrate the importance of good data and experimentation in an ever-changing business world.
To begin, Dubner pointed out that poultry consumption has vastly increased in Australia and the United States over the past half century. He wondered why, and while researching he learnt that turkey meat is the big riser and that most turkeys bred for consumption are now bred through artificial insemination. Looking deeper, he found that most chickens still breed naturally but turkey consumption habits have changed. Many people now eat turkey breast as part of their ordinary diet, whereas turkey used to be primarily consumed whole on special occasions.
In order to meet demand for turkey breast meat, poultry farmers bred turkeys to have larger breasts. Eventually their breasts became so large that they became physically incapable of breeding naturally — these big-breasted turkeys could not have sex with each other.
This story, Dubner explained, can help illustrate that you need data to understand the world. But data on its own has little use, especially today — when, contrary to 15 years ago, we don’t need more data so much as we need more people who know how to ask the right questions of the data.
In his experience working with businesses, Dubner said that “the people who are most comfortable with data in the firm are often the IT people, [but] the people who most need to unlock the mysteries of that data are in different departments”. Those two groups don’t necessarily communicate well with each other.
In the latest of our Telstra Vantage Behind the Mic series, Adam Spencer speaks to Stephen Dubner. Subscribe to the podcast via apple.
Look beyond the noise
Dubner explained further that the noisiest part of a problem is often the symptom, not the cause. Artificially-inseminated turkeys with gigantic breasts are a symptom of the agricultural revolution, which has for over a century repeatedly squashed arguments that we will run out of food to feed the world’s population. The US throws away 40 percent of food bought for consumption, he stated. The problem causing famine is not a lack of food but rather that political, social, and economic dysfunction prevents the food from reaching the people who need it.
The challenge, then, beyond asking the right questions to find the root cause of a problem, is finding the right incentives to fix the problem. Or as Dubner put it, “If you want to make a good decision, if you want to change someone’s behaviour, you have to understand the incentives that are at play.”
To illustrate that point, he next turned to bathroom hand hygiene and the difference between declared preferences and revealed preferences. He polled the audience: who doesn’t wash their hands after using the toilet? No hands went up. That’s people’s declared preference — the thing they hope to do, their expected behaviour. But his own efforts to measure this, by lingering at the sink at public toilets, have suggested that only around 70 percent of men actually wash their hands after using the toilet. That’s the revealed preference.
Hopes don’t always match behaviour, even among doctors at a hospital — who in one study self-reported hand hygiene compliance of 73 percent but were then recorded by their nurses as having an average compliance rate of only 9 percent.
Poor hand hygiene leads to bacterial infection, which is the most common cause of preventable death in hospitals. Doctors are the most educated people in a hospital, so clearly education is a poor incentive for good behaviour. Dubner talked about a committee for hand hygiene at a US hospital that tried to find the right incentive. First they hypothesised that it was a communication problem, so they issued a memo. Nothing changed. Then a subcommittee thought to use positive reinforcement — they hid in the room before a doctor arrived for his rounds, then if he washed his hands they jumped out, applauded, and handed over a $10 Starbucks gift card. Doctors loved the free gift and hurried to wash their hands whenever they heard the subcommittee was nearby.
But it didn’t change the overall rate of hand hygiene, either.
The winning solution finally came from a quiet member of the committee — “the best ideas often come from quiet people who like to sit in a room with their own thoughts” — who got everybody in the committee to press their hand into a petri dish and then got those cultured. Most returned covered in bacteria. The committee then took a photo of one of these palm-shaped cultures and used it as a screensaver on every computer in the hospital, with an explanatory label. Compliance leapt to 100 percent.
Be ready to experiment
“The moral of the story is that the incentives that you think may work don’t always work, [so you should] work hard to use data to find out what’s actually the behaviour and be willing to experiment,” said Dubner, who also pointed out that sometimes incentives can even backfire and produce the opposite behaviour — as was the case in a town that had a problem with rats eating people’s trash and spreading disease.
The government first offered free extermination, then when that didn’t work they offered free bins with lids. That also didn’t work, so they offered a cash reward. For every dead rat, they promised to pay the equivalent of about US$4. That then gave rise to a rat farming economy. Far from exterminating the rat population, people would breed rats to kill them and claim the reward.
“The next time you come up with a plan that seems perfect on paper, experiment, try it out, gather some data,” Dubner concluded, “because it may work brilliantly or you may just end up with a pile of dead rats on your doorstep.”
Sign up today for more daily articles and videos from Telstra