T Princeton University HIS paper presents an empirical analysis of the determinants of the probability that an individual will vote in Presidential elections. Our interest in this subject is prompted in part by an interest in social policy. In the last fifteen years there have been drastic changes in law, and in the interpretation of law, as it bears on voting. The Constitution has been altered by both the 24th Amendment, outlawing the poll tax, and the 26th Amendment, lowering the age of voter eligibility from 21 to 18 years of age.' The Voting Rights Act of 1965 sent federal registrars into many Southern communities that had low rates of registration and a history of discrimination against blacks, and it suspended literacy tests as a qualification for voting in six Southern states. Amendments to the Voting Rights Act in 1970 suspended such tests in all states and abolished residence requirements for voting in presidential elections. In 1972 a decision of the U.S. Supreme Court sharply reduced the length of residence that could be required for voting in state and local elections. All of these changes were calculated to enlarge the number of voters exercising the right to vote, and all except the lowering of the age of eligibility were expected to increase rates of voting. That they achieved this latter objective, even among voters 21 and older, is not obvious, the turnout of voters in the presidential election of 1972 was markedly lower than it had been a dozen years earlier. We are interested also in a better explanation of voting. Since 1960 understanding of participation in elections has also increased, even if not so dramatically as the pace of change in electoral law. Fifteen years ago solid knowledge about voting was based almost entirely on simple correlation analyses that related voting to dozens of demographic and sociopsychological variables. Since then, multivariate analyses have become more common and there have been conscious efforts to integrate theoretical and empirical analyses of voting behavior. We designed our research to build upon these trends. To a large extent, the major theories of voting have developed independently of each other; we decided to make them compete to explain a given body of data. We also