Tuesday, April 29, 2008

On Quantities



This is a post trying to spell out some baby ideas I have about a problem I have noticed in political debates. They are very hesitant ideas and don't speak yet, but I'm going to let them babble to you.

The problem is this: We use quantities, numbers, percentages, fractions and probabilities very poorly on the whole. For instance:

Is the country severely divided on some issue if 20% think strongly one way and 80% think strongly the other way? I would have thought that the country is fairly agreed on whatever that 80% think. But many disagree with me and address the beliefs of those 20% as if they have the same weight.

Now, perhaps it is the right thing to do, to treat each idea as an equally important issue. But to argue that the same number of people fall into each camp really is incorrect, and this is what I see happening in many ways.

It's just another example of the way we run many public debates: Find representatives for each extreme position and let them fight it out. The idea is that somehow one of the extremes might be right, and that the audience should be able to judge from the debate which it is. Of course this leaves out all the other standpoints but the two extremes. It also gives extreme attention to two positions that actually only a small fraction of the general population believe in. More importantly, the opinions of people in the middle are totally ignored in this setup.

A slightly different example of the quantity problem cropped up on the Eschaton comments threads one night, where people argued that killing just one person is as bad as being a mass-murderer of millions. Something that would be a fair philosophical argument turned into a very different argument, one implying that it doesn't matter how many people one murders. That the amount of suffering, pain and grieving is increased by each new victim was somehow ignored. I think that quantities do matter, and something odd is happening here, something quite frightening.

Here's a feminist example of the same trend: You discuss something that happens to women a lot (say, sexual harassment at work), and someone points out that it happens to men, too. That it happens to men, yes, but that it happens to men much, much less often is ignored. Quantities do matter.

Another example of this, but now with a very different twist: A study on gender differences shows that men do something (say, read news on the net) 60% of the time while women do that same thing 40% of the time. A difference, yes, but not a difference of 100% to 0%. But that's how it becomes, over a couple of rounds of interpretations ("Men read news on the net! Women socialize!"). It even turns into that dualistic position of the initial percentages are 63% vs. 56%.

Or think about the examples from studies about health care. If rates of breast cancer are higher among the women who have used estrogen therapies by some fairly small number the message that most people seem to take home is that estrogen therapies KILL YOU, and that somehow not getting them will save you from breast cancer.

Health care studies almost always have results which mean that there are small differences between two treatment patterns or small differences between disease prevalences in two populations and so on. But the interpretations are never like that. They are practically always dualistic: Live or Die!

This is how far my baby thoughts have gotten. I suspect that what I'm talking about is the tendency to ignore nuances altogether, combined with the tendency to go for either-or solutions which may well be part of the way humans think (Is it safe to eat? Will it kill you?). But there's also a tendency to shift the discussion from a sphere where numbers don't matter to spheres where they do matter, without us noticing that this is what just took place.