Now here is where the study gets interesting. Mullen and his colleagues then called up people in a number of cities around the country who regularly watch the evening network news and asked them who they voted for. In every case, those who watched ABC voted for Reagan in far greater numbers than those who watched CBS or NBC. In Cleveland, for example, 75 percent of ABC watchers voted Republican, versus 61.9 percent of CBS or NBC viewers. In Williamstown, Massachusetts, ABC viewers were 71.4 percent for Reagan versus 50 percent for the other two networks; in Erie, Pennsylvania, the difference was 73.7 percent to 50 percent. The subtle pro-Reagan bias in Jennings's face seems to have influenced the voting behavior of ABC viewers.

As you can imagine, ABC News disputes this study vigorously. ("It's my understanding that I'm the only social scientist to have the dubious distinction of being called a 'jackass' by Peter Jennings," says Mullen.) It is hard to believe. Instinctively, I think, most of us would probably assume that the causation runs in the opposite direction, that Reagan supporters are drawn to ABC because of Jennings's bias, not the other way around. But Mullen argues fairly convincingly that this isn't plausible. For example, on other, more obvious levels — like, for example, story selection — ABC was shown to be the network most hostile to Reagan, so it's just as easy to imagine hard-core Republicans deserting ABC news for the rival networks. And to answer the question of whether his results were simply a fluke, four years later, in the Michael Dukakis-George Bush campaign, Mullen repeated his experiment, with the exact same results. "Jennings showed more smiles when referring to the Republican candidate than the Democrat,"' Mullen said, "and again in a phone survey, viewers who watch ABC were more likely to have voted for Bush."

Here is another example of the subtleties of persuasion. A large group of students were recruited for what they were told was a market research study by a company making high-tech headphones. They were each given a headset and told that the company wanted to test to see how well they worked when the listener was in motion — dancing up and down, say, or moving his or her head. All of the students listened to songs by Linda Ronstadt and the Eagles, and then heard a radio editorial arguing that tuition at their university should be raised from its present level of $587 to $750. A third were told that while they listened to the taped radio editorial they should nod their heads vigorously up and down. The next third were told to shake their heads from side to side. The final third were the control group. They were told to keep their heads still. When they were finished, all the students were given a short questionnaire, asking them questions about the quality of the songs and the effect of the shaking. Slipped in at the end was the question the experimenters really wanted an answer to: "What do you feel would be an appropriate dollar amount for undergraduate tuition per year?"

The answers to that question are just as difficult to believe as the answers to the newscaster's poll. The students who kept their heads still were unmoved by the editorial. The tuition amount that they guessed was appropriate was $582 — or just about where tuition was already. Those who shook their heads from side to side as they listened to the editorial — even though they thought they were simply testing headset quality — disagreed strongly with the proposed increase. They wanted tuition to fall on average to $467 a year. Those who were told to nod their heads up and down, meanwhile, found the editorial very persuasive. They wanted tuition to rise, on average, to $646. The simple act of moving their heads up and down, ostensibly for another reason entirely — was sufficient to cause them to recommend a policy that would take money out of their own pockets. Somehow nodding in the end mattered as much as Peter Jennings's smiles did in the 1984 election.

There are in these two studies, I think, very important clues as to what makes someone like Tom Gau — or, for that matter, any of the Salesmen in our lives — so effective. The first is that little things can, apparently, make as much of a difference as big things. In the headphone study, the editorial had no impact on those whose heads were still. It wasn't particularly persuasive. But as soon as listeners started nodding, it became very persuasive. In the case of Jennings, Mullen says that someone's subtle signals in favor of one politician or another usually don't matter at all. But in the particular, unguarded way that people watch the news, a little bias can suddenly go a long way. "When people watch the news, they don't intentionally filter biases out, or feel they have to argue against the expression of the newscaster,'' Mullen explains. "It's not like someone saying: this is a very good candidate who deserves your vote. This isn't an obvious verbal message that we automatically dig in our heels against. It's much more subtle and for that reason much more insidious, and that much harder to insulate ourselves against."

The second implication of these studies is that nonverbal cues are as or more important than verbal cues. The subtle circumstances surrounding how we say things may matter more than what we say. Jennings, after all, wasn't injecting all kinds of pro-Reagan comments in his newscasts. In fact, as I mentioned, ABC was independently observed to have been the most hostile to Reagan. One of the conclusions of the authors of the headphones study — Gary Wells of the University of Alberta and Richard Petty of the University of Missouri — was that "television advertisements would be most effective if the visual display created repetitive vertical movement of the television viewers' heads (e.g., bouncing ball)." Simple physical movements and observations can have a profound effect on how we feel and think.

The third — and perhaps most important — implication of these studies is that persuasion often works in ways that we do not appreciate. It's not that smiles and nods are subliminal messages. They are straightforward and on the surface. It's just that they are incredibly subtle. If you asked the head nodders why they wanted tuition to increase so dramatically — tuition that would come out of their own pockets — none of them would say, because I was nodding my head while I listened to that editorial. They'd probably say that it was because they found the editorial particularly insightful or intelligent. They would attribute their attitudes to some more obvious, logical cause. Similarly the ABC viewers who voted for Reagan would never, in a thousand years, tell you that they voted that way because Peter Jennings smiled every time he mentioned the President. They'd say that it was because they liked Reagan's policies, or they thought he was doing a good job. It would never have occurred to them that they could be persuaded to reach a conclusion by something so arbitrary and seemingly insignificant as a smile or a nod from a newscaster. If we want to understand what makes someone like Tom Gau so persuasive, in other words, we have to look at much more than his obvious eloquence. We need to look at the subtle, the hidden, and the unspoken.

What happens when two people talk? That is really the basic question here, because that's the basic context in which all persuasion takes place. We know that people talk back and forth. They listen. They interrupt. They move their hands. In the case of my meeting with Tom Gau, we were sitting in a modest-size office. I was in a chair pulled up in front of his desk. I had my legs crossed and a pad and pen on my lap. I was wearing a blue shirt and black pants and a black jacket. He was sitting behind the desk in a high-backed chair. He was wearing a pair of blue suit pants and a crisply pressed white shirt and a red tie. Some of the time he leaned forward and planted his elbows in front of him. Other times he sat back in his chair and waved his hands in the air. Between us, on the blank surface of the desk, I placed my tape recorder. That's what you would have seen, if I showed you a videotape of our meeting. But if you had taken that videotape and slowed it down, until you were looking at our interaction in slices of a fraction of a second, you would have seen something quite different. You would have seen the two of us engaging in what can only be described as an elaborate and precise dance.