Tuesday, February 09, 2010

Salience & policy making

Despite all one hears about evidence-based-policy-making, its striking how people attach seemingly arbitrary significance to some events rather than others. So some deaths, for example, generate a big story and calls for action and not others. I was reminded of this in a piece below from Andrew Gelman's blog:

An article in The Guardian says;
David Champion, director of automobile testing for Consumer Reports magazine, said the core problem of faulty Toyota accelerators had been linked to 19 deaths in a decade, amounting to two a year of the 40,000 people killed annually on American roads.

"I find it a little odd that we're going to have a Congressional hearing to look at those two deaths out of 40,000," said Champion.


Liam Delaney said...

Kevin - I have thought a lot about this issue over the years. I have a lot of points on this but let me just raise one that in particular is important. One of the chief things that generates salience is the perception that an external agency may damage one as a person. Thus, I am more likely to be afraid of dying in a terrorist attack than of heart failure or dying by suicide though the latter are far more likely. We can think of many many more examples. However, I wonder whether the story that politicians given more weight to salient rather than likely events is somewhat confounded by the fact that this is what the political system is supposed to do, namely protect people's human rights from imposition by external agencies. In this context, going after a car company that may have caused two deaths a year is to protect the rights of the people involved not neccesarily to statistically minismise risk. Once again, I can think of many examples where I (even knowing the full probabilities) would still support government expenditure to minimise a risk. For example, I would support money to be spent on tracking down child murderers even though the money would potentially save more lives if placed into maternal health promotion.

This is really an issue that could merit a full conference here.

Liam Delaney said...

Sorry for typos. Another point that is being debated very frequently in the salience discussion is under what circumstances should governments try to maximise utility based upon subjective (and perhaps inaccurate) risk perceptions as opposed to objective risk perceptions. For example, in a famous Yes Minister episode, a company is producing a product called "metadioxin", which is said to be harmless. A recent story about an italian plant leaking dioxin into the atmosphere emerges and scares the hell out of everyone. The chief science advisor is absolutely outraged at the stupidity of the people involved and doubly outraged when Hacker pulls a few political strokes and gets the project killed, thus keeping everyone calm but losing lots of needed investment and jobs.

The obvious reading of the parable is that Hacker has destroyed a perfectly good project by political messing around. However, it is also plausible that he raised the utility of the wider community there who had revealed strongly their preference not to have this plant located in their area. In this case, the only reason for them not wanting the plant is that the chemical being produced sounded like something that could be harmful but is in fact not harmful.

Kevin Denny said...

Good points. One can think of good reasons for putting "excess" weight on some events. For example, many of the 40,000 road deaths p.a. in the US are just really bad luck. But you can do something about dodgy gas pedals in Toyotas. Whether that explains all the focus is unclear because there may well be easily more preventable deaths. So I suspect that there is strong element of irrationality in how we get worked up about somethings (there are also vested interests only too happy to see Toyota crash).
But I think you are suggesting a rights based approach. But in your example, while children have a right not to be murdered, don't mothers too have a right not to die prematurely? So how do you decide where to focus?
Your point about what the government should maximize is a hard one. I need some Oxygen Di-Hydride to think about that.