In social science, much research centers on the idea of preferences. Preferences are a key element of decision theory, which suggests that individuals have some order of preference over a set of outcomes. More generally, we can think of such varied topics as voter choice and grocery shopping as preferences. There are two ways to measure preferences.
The first is stated preferences. Often, individuals even have preferences over things that they cannot control or we cannot measure directly. Say, for example, that we are trying to ascertain how strongly people prefer environmentally friendly products to less eco-friendly products. If we just want to know their stated preferences, we can simply ask them. (Although, if we want to run a scientific survey, it gets less simple the more accurate we want to be.)
On the other hand, say we want their revealed preferences, so we observe their actual behavior. Behavior with regard to eco-friendly products centers on purchases, so we look at whether individuals tend to purchase eco-friendly products more than the alternatives. (It turns out they don't. h/t to Adam Ozimek) Economists generally prefer to work with revealed preferences, and as a political scientist so do I. The definitions of "hard" or "soft" data might be fuzzy, but when I talk about hard data this is the kind of thing I mean: actual, observable behavior. Sometimes survey research is your only alternative, but I have a hard time thinking of an instance in which it would be ideal. [edit: it's important to note--which I neglected to do earlier--that revealed preferences often get screwed up outside of the context of a free market; in the US even sugar isn't a free-market commodity thanks to high tariffs; this potentially undermines what I say later about health care but I think the main point still holds]
Now to apply this to health care. Karl Smith had a post over the weekend citing several studies that indicate that consumers don't actually care how effective certain hospitals/treatments are when making health care decisions. This is based on actual patient behavior (revealed preferences). Of course if you ask anyone, they would say that they want the most effective treatment possible.
This is the difference between stated and revealed preferences--everyone says they want effective treatment, but when it comes to making health care decisions, as Karl argues, often they simply feel the need to do something. On one level this is completely fine: if you're spending your own money and it's just to make you feel better in an emotional sense, go for it. Whether that means purchasing lots of movie tickets, lollipops, health care, whatever. But when you start spending other people's money (government health care) it seems reasonable that the public ("other people") take an interest in the efficacy of various procedures or treatments. I'm not talking about death committees or anything like that, but if the government is funding your health care then they have a right to say no to certain treatments. This is the same principle that leads me to believe that the government should have a right to tell people in NYC that they can't buy cola with food stamps. It doesn't mean you can't buy cola (or health care), it just means that you have to use other (ahem, your own) money to do it. If it's too expensive, find a cheaper alternative. That might mean more competitiveness in the market for health care. It might mean people going to other countries for treatment (not ideal but certainly possible). And it might even mean that more people become aware of homeopathic alternatives.
On a lighter note, here's a clip from Scrubs that always comes to mind when I think about the effectiveness vs cost of certain health care procedures:>