Here's how it works: Let's say that a contestant on Who Wants to Be a Millionaire uses the audience-poll lifeline, on which the audience is almost always right. (This seems odd to some, that an audience of people who would show up to watch a TV show being taped would be right so often, but the outcome is skewed by the fact that most contestants are smart enough to consult the audience strictly on pop-culture questions. The audience is far less likely to be overwhelmingly correct on science or geography questions.)
Let's assume that a contestant asks a question that a majority of the audience will get right. However, if the contestant asked that same question to the same audience, but the people had to respond verbally, one after another, it's almost a mathematical certainty that the outcome would be different (in one direction or another).
Suppose the first person gets it wrong. If the second person isn't absolutely certain of the answer, he'll probably go along with the first person. The third person is even less likely to buck what now appears to be a trend, and so on. After a while, it's going to take someone who is absolutely certain of the answer to go in a different direction, and even then, how many times have you been sure of the right answer and felt a twinge of doubt, because everybody else was taking a different route?
Compounding the matter is a myriad of factors. What if the first person polled responds quickly and forcefully? What if he has a deep voice? What if he/she wears glasses? Hesitates before answering? Is Asian? Fidgets? "Looks" smart? Is tall? Changes his answer? Any one of these things can send the group cascading off in one direction or another. Indeed, the number of factors that have at least a small effect on the outcome can be almost limitless, and the resulting permutations can be infinite.
A study of group dynamics can be useful. Research has shown that groups are surprisingly susceptible to reaching the wrong conclusion, even when a majority of the individuals know better. One such study gave 60 percent of group members information that would lead them toward the correct answer, while the rest were given information that would lead them the other way. In such cases, there is still a 1-in-3 chance that the group will cascade toward the wrong conclusion.
The results are further skewed when people start with (false) preconceived notions and/or strongly held (but false) beliefs. Take this basic statement: There is a direct correlation between the amount of fat in a person's diet and the risk of heart disease. Pretty basic, right? The only trouble is that it's not true. Never has been. It might seem like it should be true, but countless clinical trials have failed to establish a link. What has happened over the past few decades is that organizations like the U.S. Department of Agriculture (with the "food pyramid"), along with certain members of the media (who lazily prefer clear, simple statements over correct ambiguity), have cascaded over one another until the fat intake-heart disease "link" became chiseled in stone.
It's even more pronounced when zealots are allowed into the equation. For example, just try to find a vegetarian who doesn't believe that he's going to live longer than his omnivorous counterpart. The study that many point to in order to back up that dubious contention concluded that Seventh-day Adventists (who are vegetarians) live about four years longer than do people who don't follow that particular faith. The only problem is that the study failed to take into account the fact that Seventh-day Adventists also don't drink or smoke and generally don't partake in high-risk behavior, such as having multiple sex partners. Factor in those points, and the life-expectancy difference goes away.
Even if they were to cherry-pick results from different studies, the best that a hopeful vegetarian could point to is an increased lifespan measured in months, a figure that, during an average lifespan of 75 years, is statistically negligible. And yet, among members of that group, the cascade toward a false conclusion is almost universal.
Clearly, the most destructive example of cascading is that which led to George Bush's invasion of Iraq. Think back to how one false piece of information led to another and established a base from which still others would spring. The fact that the percentage of Americans who still believe that Saddam Hussein had something to do with Sept. 11 is greater than zero points to cascading (and some really dumb Americans).
The proliferation of "news" outlets, coupled with the blogosphere on the Internet and the ever-shrinking attention span of the average American, pretty much guarantees that cascading is going to get worse. About the best we can hope for is that people come to recognize that the phenomenon exists, and this leads to an understanding that one doesn't have to be a part of the problem.