When we are choosing which action to take, one of the most basic calculations which guide us is, "How likely is it to lead to one option or another." We need to think of all possible outcomes, and the rough probability of each one occurring. There is a problem with this. We are not great at assessing probability. But there are certain ways to make us think that one outcome is more probable than another.
Add a little description, especially when that description fits in with predetermined pictures, and people will chuck probability out the window. For example, give most people a choice between "Bob owns three guns" and "Bob owns three guns and hunts deer," and they will likely choose the latter as the more likely option. This can't possibly be true. Both Bobs own three guns, and imposing more conditions on one of them only makes their situation less probable. But it fits in with a stereotype of gun owners, and so people believe it "more."
As it turns out, this doesn't just work when applied to people. It also works when applied to situations. When Paula Jones brought a lawsuit against sitting President Clinton, it didn't take long for speculation to start about how it would turn out. A group of 200 lawyers was asked to predict how the trial would end; some were asked basic stuff, like would the trial reach a conclusion, or would it be stopped prematurely for some reason. Others were asked to break down a list of reasons why it might be stopped prematurely, and then to predict whether the trial would go to conclusion. The lawyers who got a list of possible reasons why the trial would be stopped thought that a premature end to the trial was much more likely than the lawyers who were only asked if the trial would reach a conclusion or if it would not.
This isn't clinching proof, of course. The lawyers who received a list, being reminded of all the ways the trial could be stopped, might have had a better estimate as to how the trial would go. But it happens with doctors as well. If a disease has multiple symptoms, some common and some rare, surveyed doctors believed that the rare symptom alone was less likely than the rare symptom and the common one combined. When looking at pure probability, more factors mean something is less likely, but we immediately think of more factors being more likely.
In the end, what we want is a vivid description. If we can see Bob and his guns, or the many outcomes of a trial, or a patient displaying all the symptoms of a disease, the idea attains a reality for us. Or for whoever we're describing it too. This means that, if we want to manipulate people, all we need to do is present them with many choices, some spare but likely, others richly detailed but unlikely. Even as we pile idea on idea, making the probability less and less likely, it will seem more and more likely to whoever we're manipulating. They can make a free choice, but it's biased in our favor.
[Via The Drunkard's Walk]