Second-guessing poll position

Americans cast their votes as individuals, but presidents are elected by winning a majority in the electoral college. In the language of TV politics, red states are those that cast their electoral college vote for a Republican, and blue states are those that vote Democratic. As Al Gore learned in 2000, you can get more votes than your opponent but still lose the White House because the runner-up in the popular vote can become president with a majority of the electoral college vote.

The thesis of this topical book is that how Americans vote depends on where they live as well as who they are. Andrew Gelman, a professor of statistics at Columbia University, makes this argument clearly and repeatedly in colloquial language, in black-and-white graphics and in maps coloured red and blue.

The underlying assumption is familiar to a British audience: class counts. Lower-income voters ought to vote Democratic, and upper-income voters favour the Republicans. However, there is a puzzle. Today, lower-income states in the American South such as Mississippi are Republican and upper-income states such as Massachusetts tend to vote Democratic.

This paradox is familiar to anyone who has canvassed for Conservative votes in North Oxford or for Labour votes in Glasgow East at this summer's by-election.

Gelman ascribes this difference to an ecological effect: people vote not only according to their individual characteristics, but also in response to social pressures from those around them. The conclusion is demonstrated by analysing survey data showing differences in class divisions within states. As one might expect, better-off people are more likely to be Republican in Mississippi, making that state go red; in contrast, in Massachusetts better-off people are more likely to support the Democrats. In Britain, a similar phenomenon has occurred: middle-class voters in South Wales have tended to be more Labour than their counterparts to the east of the Bristol Channel. It is also familiar in a Labour-controlled Downing Street, where capturing the Daily Mail vote is a priority and the shrinking working-class vote has been taken for granted.

A major strength of the book is that it shows the importance of changes in America in the past half century. Race, not class, was critical in creating the electoral geography of today. After Lyndon Johnson identified the Democratic Party with the civil rights cause, poor Southern blacks got the right to vote and supported the Democratic Party. However, this drove poor Southern whites into the Republican camp. Religion complicates the picture further, for poor whites as well as blacks are more likely to be born-again Christians. At this point, the author's fixation with two-dimensional maps and graphs becomes a constraint. They cannot show the multiplicity of influences on voting that can be identified by tables of multi-variate statistics.

The surprises of US politics limit the value of this book for predicting the outcome of the 2008 presidential election. The choice of a woman moose-killer as Republican vice-presidential candidate is novel, and the polychromatic character of Barack Obama makes him different from any previous presidential candidate in American history. The big unknown is whether Obama's candidacy will, like Al Smith's defeat in 1928, show that a Catholic (read: black) cannot be elected president or, like John F. Kennedy's victory in 1960, show that a Catholic (black) can be elected president, albeit with less than half the popular vote but with a majority in the electoral college delivered by the state of Illinois.