Simply rolling out an untested idea often has fewer roadblocks than researching whether it works, but that doesn’t make it ethical. Behavioural insights experts in government discuss the do’s and don’t’s of their research.
As one of the founders of Harvard’s Behavioural Insights Group and now the head of the Department of the Prime Minister and Cabinet’s nudge unit, Professor Michael Hiscox says he doesn’t have a lot of time for fairness concerns around randomised control trials in policymaking. He’s heard it all before.
Asked about the most tedious common argument against randomised trials at an Institute of Public Administration event in Melbourne last week, Hiscox responded that “the first kneejerk response is something along the lines of ‘this can’t be ethical’.”” … we have to clear a much higher ethical hurdle to see if works than if you’d just done it without research.”
His response is to explain that this is how medicine is tested — but also to ask, “isn’t the first order ethical problem do you know if you’re going to cause more good than harm with this policy that you think is going to work? And how do you know?”
In fact, it’s a strange quirk of the system that while differential policy testing is seen to be ethically dubious, rolling out a new policy untested doesn’t attract the same kind of response.
“Isn’t it ironic that in order to do research to figure out whether this policy works we have to clear a much higher ethical hurdle than if you’d just done it without research? Which is kind of extraordinary,” Hiscox noted.
“You can implement a massive social or economic policy at scale without knowing whether it will work. If you want to do a trial to test whether or not you’re going to do harm, well wait that can’t be ethical, you’ve got to slow down and have some extra things in place before you do that.”
‘You’ll die’ vs ‘you can win an iPad’
David Halpern, CEO of the UK Cabinet Office’s Behavioural Insights Team, believes “it’s worth being respectful of the concerns — not least because that’s likely to be more effective when the argument arises.”
But while sceptics tend to worry about the potential for new initiatives to cause harm, governments often cannot say for sure the current system is not doing the same.
The best counter-argument to questions about whether it is fair to treat people differently in the name of a trial, he says, is the “sobering thought” that “most of us who work in this area have a whole list in our minds of things people thought were really good ideas, often we’d been doing for years and years and years, that turned out to be not just not effective, but often counterproductive”.
When people are confronted with those types of examples, “it is pretty compelling to then say, are you sure not only of the thing you’re about to do, are you even sure if the thing you’re doing right now is effective?” Halpern argues.
And while the whole basis of behavioural economics is that humans are often irrational, and often in predictable ways, Halpern explained that this holds true when it comes to questions of life and death.
He recalled one adviser’s astonishment when he outlined how just telling people their life is at stake is often not very useful. His response was: “let’s get this straight — if the person doesn’t change, they will die. You want to talk about other kinds of incentives?!”“Even if tests show an intervention doesn’t work … making the information available stops others having to repeat the same testing.”
But, Halpern says, “it’s not a very effective strategy to say, if you don’t do this, you’ll die, as opposed to saying we’ll put you into the lottery for an iPad.
The same goes for things like climate change. “You’re not going to save the planet on a rational argument, sadly, for the most part. You may be able to get people to insulate their homes through other routes,” Halpern argued.
BI team gets top-level oversight
The Commonwealth government’s Behavioural Economics Team, run by Hiscox, was launched on February 1 as a joint initiative of 13 agencies and hosted in PM&C.
Apart from examining interventions in a large range of policy areas — Hiscox was reticent to say exactly what they’d be working on before more details are ironed out in coming months — its work will include looking at internal processes such as human resources, as well as training up public servants in behavioural insights and assisting agencies that haven’t already built up their own BI capability.
The board will be chaired by either Martin Parkinson or David Gruen, said Hiscox. At the moment they’re also hiring lots of researchers and advisers.
A central policy randomised control trial register is on the agenda too, he revealed, who is a big advocate of pre-registering trials. Even if tests show an intervention doesn’t work — results that are often hidden to avoid embarrassment at the moment — making the information available stops others having to repeat the same testing, he said.