Sunday, March 04, 2018

Meet the MP who actually wants to know what works

A house in the Canberra suburb of Narrabundah had suffered multiple burglaries. For the sixth time someone had broken in through a window and taken things from the bedroom of a nine-year-old boy.

This time the offender was caught red-handed, with a pillowslip full of the Lego – he was the nine-year-old boy from next door.

When police officer Rudi Lammers arrived, he decided not to follow the usual procedure. Instead, he sat down with the two nine-year-olds and asked the victim, “What do you think we should do?”

The reply surprised him. The victim tipped out half the Lego, and gave the rest to the thief. Then he said, “Any time you want to play Lego, come over. But can you come through the front door? Because Dad gets really cranky when you come through the window.”

Decades later, Lammers was approached by a man in a Canberra club who whispered in his ear, “Do you know who I am? I am the Lego boy. That experience changed my life.” He had stopped stealing and gone on to run a construction company.

Without realising it, Lammers had been practising what’s known as ''restorative justice''. We know that it works better than the other kind, not just because of that incident (''data'' is not the plural of ''anecdote'', Labor politician Andrew Leigh reminds us in the new book that recounts the story) but because restorative justice has been trialled repeatedly in random experiments.

Whenever offenders have been randomly assigned either to arrest or meeting the victims, those assigned to meeting the victims have been significantly less likely to commit further crimes, even violent crimes.

Except in the family.

In the Minneapolis domestic violence experiment, police were given a special pad of report forms which listed three kinds of responses: arrest, send the perpetrator away from the home for eight hours, or provide advice. The three responses appeared in random order throughout the report pad. Where the injury wasn’t serious, the police had to do what the pad said.

The results were crystal clear. Households where the perpetrator had been arrested experienced half as much violence in the following six months as those in which the perpetrator had been temporarily removed or given advice. Arrest became the standard response.

Originally an economics professor, Leigh might well have the least number of preconceptions of anyone who has ever sat in parliament. On teenage sex, he reports that virginity pledges don’t work. On smoking, he reports that paying people to stop does work. Vitamin tablets and fish oil are of no use to healthy people, so Leigh has given them up. None of these findings was obvious, and none would have become apparent were it not for random trials, which Leigh wants to use when in government.

Rather than introduce co-payments for bulk-billed visits to the doctor, as Tony Abbott wanted to do, we could introduce them in one state, or postcode, and see how over-servicing and under-servicing or visits to emergency departments in that location change. Or we could make use of the giant Rand health experiment, in which Americans were randomly assigned to free visits, cheap visits and expensive visits. It found that co-payments cut visits, but that they do it indiscriminately, keeping away genuinely ill people just as much as the not so ill.

Advertisers do it all the time, running test campaigns in small states before going national. Pharmaceutical companies do it. By experimenting with different colourings in random trials they know that yellow tablets work best at fighting depression, white ones at reducing pain, green ones at fighting anxiety and blue ones at sedation.

Surgeons do it. But to ensure the experiments make sense, they have to actually perform surgery - sham surgery - on knees and sometimes brains, where people get put under anesthetic, cut open, not operated on, and stitched back together again. Three-quarters of the time those who get sham surgery improve just as well as those who get the real thing.

There are ethical problems, to be sure. But they are often overstated. My economics lecturer told me of an early researcher who reported the results of a study in which he had withheld treatment from randomly chosen patients. He displayed a graph that showed those who missed out died sooner. Amid uproar and suggestions he had blood on his hands, he apologised to the conference and turned the graph the right way around. By withholding treatment he had prolonged their lives.

Politicians are forever promising programs they say are ''evidence-based'', but they are usually less than keen on conducting the experiments needed to obtain that evidence. Leigh is different, which means that in government he’ll probably be different. He used randomly-generated advertisements to pick the title of the book. Randomistas: How Radical Researchers Changed our World got the most clicks.

In The Age and Sydney Morning Herald