National Election Pool (NEP)

The National Election Pool (NEP) is a consortium of news organizations—ABC, the Associated Press (AP), CBS, CNN, FOX, and NBC—that conducts exit polls, related surveys of voters, and samples of tabulated vote in U.S. elections. These data allow NEP members to project or “call” winners of many political races earlier than would be possible based on final vote count alone. The voter surveys also allow pool members and subscribers to analyze demographic, attitudinal, and other variables that help explain election outcomes.

Typically the exit polls and sample vote count cover top-of-the-ticket statewide races including those for president, U.S. Senate, and governor, as well as selected ballot initiatives. NEP also conducts a national voter survey in general elections. The NEP exit polls are among the largest one—day survey research undertakings anywhere; in the November [Page 498]2004 elections, approximately 150,000 interviews were conducted in 1,469 U.S. precincts nationwide.

NEP's roots date to 1990. Before then, several television networks fielded their own exit polls and vote count samples individually. In 1990, the broadcast networks ABC, CBS, NBC, and the then—new cable network CNN formed Voter Research & Surveys (VRS) to pool these functions. In 1993, those networks and the Associated Press, a global news network serving newspapers, broadcasters, and more recently online customers, created the Voter News Service (VNS), which merged the VRS exit polling and sample precinct vote count with the National Election Service (NES), a consortium of news organizations that tabulated vote comprehensively on election nights. The cable network, FOX News Channel, joined VNS after the 1996 presidential primaries.

Exit polls are face-to-face surveys of voters as they exit polling places on Election Day. From the time the polls open until about an hour before they close on Election Day, interviewers approach respondents at a systematic interval and ask them to complete self—administered paper questionnaires, which are kept confidential. Samples of voting precincts—stratified by geography and past vote by party—are selected for the exit polls to be representative of the state, or in a national survey, the entire country. In addition to the exit poll sample, a “superset” random sample of precincts is drawn and news stringers (part—time and/or temporary employees) assigned to report vote count as quickly as possible after polls close. As early and absentee voting began to become more widespread in the United States, VNS started supplementing some exit polls with random—digit dial telephone polling the week before the election to reach voters who would not be covered in the Election Day in-person surveys, and these data are incorporated into projections models and analytical survey cross-tabulations.

In the 2000 general election, VNS and its members became enmeshed in controversy over erroneous or premature calls in the presidential race in several states, particularly in Florida—Both early on Election Night, based in part on faulty interpretation of the exit polls, and early the next morning, based on faulty interpretation of the vote count models alone. In a congressional hearing in 2001, the VNS partners vowed to improve their systems, and subsequently they hired a contractor to do so, but the computer overhaul failed in the 2002 midterm election and no exit poll or sample precinct vote data were available that night. Thereafter, the VNS members disbanded that organization and formed NEP in its place.

...