Bennett is confused. That we advocate voter registration reform does not mean we are single-factor analysts. Bennett does not understand the distinction between our analysis of the complex causes of low turnout, on the one hand, and our analysis of the limited options for reform, on the other.

As to the first, we wrote Why Americans Don't Vote because we thought the “party behavior,” “legal institutional,” and “social-psychological” schools were each inadequate, taken alone, as explanations of low turnout. The party system explanation rests too much of its historical account on the impact of the single election of 1896 in stifling party competition and increasing internal oligarchy and fragmentation, and it is evasive on the question of why the restoration of party competition in the 20th century did not raise turnout in 19th century levels. Legal-institutional explanations fail to provide an account of the political origins of legal arrangements, and of the uneven implementation of these rules by party organizations. And social-psychological explanations do not confront the anomaly that correlations between social traits and non-voting are unique to the contemporary United States. In fact, turnout was high in the 19th century when education levels were low, and then fell in the 20th century as education levels rose.

Piven and Cloward are right. Someone is confused. Readers of PS can decide who. Piven and Cloward claim they have been misunderstood. They say their goal, particularly in Why Americans Don't Vote, was to highlight the inadequacies of three “schools” of turnout studies, “party behavior,” “legal-institutional,” and “social-psychological,” and to suggest a “synthesis” of those approaches.

“Misunderstanding,” it seems, is widespread. At least two reviews of their book (Wattenberg, 1989; Pomper, 1989) faulted Piven and Cloward for failing to explain why, if personal registration is the main reason for low turnouts, voting has fallen during a period of liberalized statutes and for a “quick dismissal of attitudinal variables” (Wattenberg). Pomper (p. 1398) noted that Piven and Cloward “contemptuously dismiss (socialpsychological) explanations as ‘normal science’ and ‘the witless use of multiple correlational analyses,’ asserting instead that ‘the political context determines whether these factors will have a significant effect on participation’ ([Piven and Cloward, 1988b:] 117–120).” Bastedo (1989) interpreted Piven and Coward's 1989 article as strictly an attack on registration laws.

The terms are not as Cloward and Piven have stated them, between one theory that holds that fewer and fewer people are registered to vote and another that holds that registered voters are going to the polls less and less, but rather whether registration barriers or lack of motivation is the primary cause of nonparticipation. It is also about the public policy implications of that debate—whether one engages in a unitary attack on registration barriers or a multi-faceted approach to nonparticipation, which includes but is not limited to barrier removal.

In the inexact world of voting statistics nothing will be fully solved without a full-validation study of Census-reported registration and voting figures, and state aggregate data on registration.

There are three issues involved: whether the state official registration rolls are inflated; to what degree; and whether they are progressively inflated. On the first there is broad agreement, which includes myself and the disclaimer I issue in each of the Committee for the Study of the American Electorate's reports on registration and turnout.

Over the last forty years the concept of swing has transformed the interpretation of elections, first in Britain and later in Australia, India and other countries. But Americans have been oddly reluctant to apply this most useful of simplifying formulas to election reporting or to the analysis of voting trends.

Swing is a device for reducing to a single statistic all the complex movements in party support between one election and another. It offers a quick and easily comprehended indication of trends and of the extent to which voting behavior differs from one area to another.

On election night in Britain, as soon as the broadcasters announce the first constituency counts, they also report the local swing, the net percentage movement between Conservative and Labour since the previous election, and on this basis they suggest what will be the final outcome ‘if the swing in the remaining seats resembles those already declared.’ Although this health warning is dutifully repeated, in fact the projection from the early results has always given a remarkably accurate pointer to the overall division of seats.