Exit Frowning

Local officials were still sharpening pencils to tally results on election night when the networks started calling winners. ABC's analysts won the rush to judgment: they were first with projections in 33 races for Senator or Governor, vs. 17 for NBC and 16 for CBS. A crucial tool for the networks was the "exit poll" of randomly chosen voters at some ten to 75 precincts per state, for a total of up to 36,000 people per network. The technique was pioneered by CBS Pollster Warren Mitofsky in 1967; since the 1980 primaries it has been a favorite statistical gizmo of all three networks. Voters are handed anonymous questionnaires asking their age, sex, race and political attitudes, as well as which candidates they chose. Exit polling can be intrusive: officials in the state of Washington alleged that CBS aides in Black Lake (pop. 15) and NBC operatives in Bellingham (pop. 45,800) had come right into polling places to solicit opinions.

Exit polling has statistical pitfalls as well. Admits a network official: "You are counting how people said they voted, not how they actually did. Moreover, the older a person gets, the more likely he or she is to vote, but the less likely to participate in an exit poll." Adds another network vote counter: "You cannot mathematically calculate the sampling error in an exit poll." Once results are in, in order to use them a pollster must guess the eventual percentages of turnout by age, race, ideology and so on. Warns I.A. Lewis, whose exit poll for the Los Angeles Times was 10 points off on both of California's major races: "Exit polls are tools to analyze. It is foolhardy to base projections on them."

This file is automatically generated by a robot program, so viewer discretion is required.