PREELECTION surveys have generated much publicity as the deadline for the filing of candidacy draws near. And, while few have formally announced their candidacies next year, a number of presidential aspirants have said that they will make their decision whether to run or not on the basis of such surveys.
Preelection polling generally makes accurate prediction and may be a factor in an aspirant’s decision. But that’s only one of several factors. And, in this Internet era, despite measurement of voters’ sentiments that employs state-of-the-art devices, the polls can be dead wrong. For instance, the recent general elections in the UK. Preelection surveys showed a hung parliament was certain, given the two major parties—the Conservatives (Tories) and the Labor Party—were running neck- and-neck race.
And even the exit polls did not read the Tories would win a resounding victory. Or that a surge of nationalism would allow the Scottish National Party to take all but three of Scotland’s 59 constituencies.
Such was the failure of the UK electoral polls that the British Polling Council—the pollsters’ industry association—conducted an independent inquiry into the matter. And, while many explanations were proposed, the most popular was also the simplest—respondents simply said one thing and did another.
The UK experience, however, does not appear to be entirely novel. Months before, during Israel’s parliamentary elections, Prime Minister Benjamin Netanyahu and his Likud party enjoyed a five-seat lead over opposition leader Isaac Herzog and the Zionist Union, whom opinion polls had predicted to win a four-seat lead in a runup vote.
Up to a year before the 1948 US presidential elections, opinion polls clearly pointed to New York Gov.and Republican candidate Thomas E. Dewey winning over Democrat incumbent Harry S. Truman. Dewey’s triumph almost became conventional wisdom. Yet the outcome confounded all, as Truman won and served another term as US president.
Methodologies and statistical techniques have advanced greatly since the 1940s and have become useful tools for making decisions.
Nonetheless, poll results should not be seen or taken as foregone conclusions, despite higher levels of confidence, larger sample sizes and slimmer margins of error. For like any other human behavior, elections are complex and, hence, unpredictable human phenomena.
E-mail: angara.ed@gmail.com.