(Editor?s Note: This article is the first installment in a series on Election Surveys 101.)

MANILA, Philippines?Surveys or opinion polls are not free from risks.

Poll results that are several weeks or months old may be perfectly valid. But surveys are snapshots of opinions. A mood or opinion is, of course, subject to change. That?s why surveys taken at different times reflect different opinions.

Survey pioneers learned this lesson the hard way.

George Gallup Jr. wrote: ?If the 1936 election performance of scientific polls gave the fledgling industry considerable credibility with the US public, their performance in the 1948 election threatened to undo everything.?

The three American polling organizations?Gallup, Roper and Crossley?were lulled into thinking that few votes would change after the start of the presidential campaign. They stopped interviewing several weeks before the presidential election and predicted a win for Republican Thomas E. Dewey.

Famously incorrect headline

When they ended their surveys early, they missed the swing of third-party voters back to Harry S. Truman?s camp in the final days of the campaign for the 1948 presidential election.

?Dewey Defeats Truman? was the famously incorrect banner headline on the front page of the Chicago Tribune on Nov. 3, 1948. It was based on an analysis that used the almost-unanimous survey results that said a Dewey presidency was ?inevitable.?

Nowadays, for the presidential election, surveys are typically conducted every month, with the frequency increasing the closer we get to Election Day.

SWS? mistake

In the Philippines, Social Weather Stations (SWS) committed a huge mistake in the conduct of exit polls in 2004.

Exit polls, which originated in the United States, are post-election surveys conducted as voters leave precincts to determine the likely winner soon after the voting.

Often, the respondents are asked to fill out questionnaires asking for their age, sex and the candidates they voted for and why. SWS was the first to conduct a nationwide exit poll in 1995. It also conducted nationwide exit polls in 1998 and 2001.

In all three exit polls, SWS had been accurate. But in its 2004 exit poll, SWS erred in declaring that President Gloria Macapagal-Arroyo had won in Metro Manila and Southern Mindanao when it was opposition candidate Fernando Poe Jr. who carried these regions.

SWS also erred in saying that Poe had won in the Autonomous Region in Muslim Mindanao (ARMM) when the official count showed otherwise.

Nonresponses

A seven-member independent committee, tasked with reviewing the conduct of the SWS exit poll and identifying the factors that led to the erroneous results, concluded that the errors were due to the high number of nonresponses and the understated margin of error used in the survey.

SWS said it had expected that 10,000 voters would be interviewed but only 4,445 were actually polled. While this number was enough for a national survey, this was ?disappointingly below expectations? when broken down into regions.

Unlike in the US exit polls, field workers for SWS conduct the exit poll surveys in the privacy of the respondents? homes. Most of the target respondents were not yet home due to the heavy rains on Election Day.

Due to the tight election race, many were also said to have stayed longer at the precincts to watch the counting.

Too close to call

Had SWS used a margin of error of plus or minus 2.8 percentage points instead of the plus or minus 1.4 percentage points, the 2004 presidential election could be described as ?too close to call, a near statistical dead heat, and not a clear victory for Arroyo,? a veteran statistician said.

The SWS 2004 exit poll shows that even pollsters with a good track record can err.

The US-based National Council on Public Polls (NCPP) says that no matter how good the survey, no matter how wide the margin, no matter how big the sample, a pre-election poll does not show that one candidate has the race ?locked up.?

Things change, often and dramatically, in politics. That?s why candidates campaign, says the NCPP.

Sources: Inquirer Archives, http://www.ncpp.org/?q=node/4, http://www.casro.org/media/History%20of%20Research.pdf, SWS website