Features unwavering in one thing: that their move- ment was patriotic, not subversive. Dis- sent was legitimate and appropriate, op- position to political leaders and policies fully compatible with the constitution of the People's Republic. The premier how- ever had another view: the students were laying the foundation of opposition parties, he said, and this was something the "peo- ple's democratic dictatorship" could not tolerate. About the Author Stephen L. Elkin is professor of political sci- ence at the University of Maryland. Editor's Note: Cloward and Piven discuss their plan for voter registration reform in the Fall 1988 issue of PS in an essay entitled, "National Voter Registration Reform: How It Might Be Won." Other political scientists have also taken up the issue of registration reform, among them Raymond Wol finger of the University of California, Berkeley. Government Statistics and Conflicting Explanations of Nonvoting Frances Fox Piven City University of New York Richard A. Cloward Columbia University When turnout fell to a near-historic low in 1988, the two sources of government statistics suggested opposite reasons. One is that fewer and fewer people are regis- tered to vote, and the other that regis- tered voters are going to the polls less and less. According to the figures supplied by state election officials who tally up county registration totals and record election re- turns, voting by registered voters is down 15 percentage points since I960: 85.4% voted that year, but only 70.5% in 1988. By contrast, the U.S. Census reported that 86.2% of registered voters cast ballots in 1988, down only 4.8 percentage points since 1968 (when the Census post- election sample surveys of both registra- tion and voting were first undertaken) (U.S. Bureau of Census, 1989). The 1988 result confirmed the earlier Census sum- mary conclusion for the presidential elec- tions between 1968-80 that registered voters "overwhelmingly go to the polls" (U.S. Bureau of Census, 1984a). The Census surveys show that falling registration is the more important cause of declining turnout: the level slipped from 74.3% in 1968 to 66.6% in 1988, or 7.7 percentage points. Adjusting that result downward by about 10% to correct for the tendency by some respondents to tell the Census interviewers they are regis- tered when they are not, the 1988 level of registration was 59%.' State data, by con- trast, show a much higher registration level: 70.9% in 1988, off only 4.5 percent- age points since I960. These differences between government sources undergird the dispute as to whether it is declining turnout by registrants or declining registra- tion levels that is the larger reason for the falling vote. On most matters, academics consider Census data superior, including studies of registration and voting. But the public is exposed mainly to state statistics. These data get the big play in the press for the simple reason that they are available at election time when the question of why turnout is falling is national news. The Cen- sus results do not become available until four or five months later, when the press no longer much cares. Consequently, the frontpage story after every election is that ever-enlarging proportions of registered voters are not showing up at the polls. In an editorial based on state data anticipat- ing the 1988 turnout drop, the New York Times said that ' The percentage of regis- tered voters who do vote has steadily de- clined—from 85.3% in I960 to 80.5% in 1968 to 72.6% in 1984" (emphasis in original) (Nov. 6, 1988). And, indeed, the state statistics available the day after the election showed that a new high of absten- 580 PS: Political Science & Politics