Behind the fascination some of us have for politics lies an enduring mystery: What is going on in the voters' heads? Why do so many states split their Senate votes between the two parties and elect such liberal-conservative opposites as Michigan's Carl Levin and Spence Abraham or Iowa's Tom Harkin and Chuck Grassley?

And what explains our mixed-up views about government--the appeal of Ronald Reagan's conservative rhetoric and the rejection of Newt Gingrich's Contract with America? How can Americans applaud Bill Clinton for saying "the era of big government is over" and then back him when he vetoes a tax cut that would actually force Washington to retrench?

An arresting new effort at decoding these mysteries can be found in "Reading Mixed Signals: Ambivalence in American Public Opinion about Government," by Albert H. Cantril and Susan Davis Cantril (Woodrow Wilson Center Press).

Their analysis is based on a survey of voters probing two subjects: first, citizens' general attitude toward government and, second, their views about specific government activities. On the general question--whether government does too many things people could do for themselves, whether it has too much power and whether more government regulation of business is needed--there is substantial division. But almost half think the government does too much and has too much power, and roughly one-third think there is too much regulation of business.

But when the scholars asked about specific programs, there was heavy sentiment that the government should do as much as or more than it is doing now. Spending on these programs has been a matter of serious dispute in Congress--job training, medical research, subsidizing teachers, college student aid, clean air standards, Head Start, job-safety regulation, Medicaid and housing assistance. Except for the last two programs, fewer than 10 percent of those interviewed wanted government to do or spend less. On the first six, half or more said government should do more.

From these results, the Cantrils draw a four-way diagram of the public. The Steady Supporters, who have a positive view of government in general and favor almost all these activities, make up 39 percent. The Steady Critics, who have a negative view of government and oppose many of the specifics, are 18 percent of the public. Ambivalent Supporters are 12 percent--yes, in general; no, in specifics. And the most interesting group--20 percent of the total--are the Ambivalent Critics, the people who are negative toward government but support most of the specific programs.

"The truly competitive arena" of politics "is among Ambivalent Critics," the Cantrils write. More than any other group, they tend to be ticket-splitters. Conservatives like Reagan appealed to them "at the general, symbolic level," but when Gingrich began to target specific government programs that they valued, he lost their support.

More liberal or activist politicians like Clinton succeed with these Ambivalent Critics when "they speak of the value of the specific things government does or should do." But these politicians are vulnerable if they appear to be defending government as a good thing in itself.

The Cantrils' manuscript was completed before Texas Gov. George W. Bush defined his approach as "compassionate conservatism." But that label clearly allows him to position himself as a critic of big government but still defend a federal role in education--as he has done. If the Cantrils are right, he may have hit on a winning formula.

But beyond these specific applications, this volume--rich in insights far beyond what can be summarized here--imparts one lesson no politician or political reporter should ever forget. People think in more than one dimension. It is far too easy to categorize voters in simplistic liberal or conservative camps or to create stereotypes like "soccer moms" or "angry white males" that conceal more than they reveal.

The Cantrils' concluding paragraph serves as a rebuke to all such thinking. Referring to both the Ambivalent Critics and the smaller group of Ambivalent Supporters, they write: "A politically important one-third of the public are ambivalent in their thinking about government. Their general views about government are often at odds with what they think government should actually be doing. These are people whose views on issues are nuanced and cannot be easily pigeonholed. Ambivalence about government appears to be integral to the way they see the political world."

The Ambivalents "hold the balance of power in American politics."