Aside from the usual mud-slinging (what's Philippine elections withut it?), the other thing that makes the forthcoming polls exciting or exasperating (depending upon whioch candidate's siad you are) are the various survey results.
Here's something I got from Philippine Business online magazine. It's an artricle on surveys in line with the 2004 elections. I guess it's contents are still relevant to the forthcoming elections (yup, the PCOS flimflam notwithsanding).
7 Rules for Reading Surveys
Heading into the 2004 elections, expect more surveys to pop up in the press. The most-often quoted ones will be those of Social Weather Stations, Pulse Asia, and ASW Roper. Don’t believe everything you read about a survey in the media. Reporters tend to gloss over the technical details of survey preparation and rush for headline-grabbing news. Here’s a quick guide on how to read surveys:
Rule 1: Check who’s conducting the survey. All surveys and polling companies are not alike. Some use scientific methods while others don’t. Make sure you are reading the results of a reputable pollster.
Rule 2: Check who’s sponsoring the survey. In the Philippines, political surveys are typically sponsored or commissioned by third parties. For instance, SWS surveys on presidential candidates are commissioned by media companies or political consultancies. The sponsors, and not SWS, submit the names to be tested in the surveys.
Rule 3: Surveys are snapshots of opinion. Surveys provide a picture of public opinion at given points in time. The mood or opinion are of course subject to change. That’s why surveys taken at different times may reflect different opinions. For presidential elections, surveys are typically conducted every month, with the frequency increasing the closer we get to elections.
Rule 4: Sampling matters. The only reliable surveys use a sampling method known as stratified, random sampling. Respondents to the survey are typically drawn randomly from all socioeconomic classes (i.e., AB, C, DE) and all regions (Metro Manila; the rest of Luzon; Visayas; and Mindanao). For a presidential survey to make any sense, all respondents should be aged 18 and above and be registered voters. Both SWS and Pulse Asia use sample sizes of 1000 to 1200 persons. All respondents are given face-to-face interviews. ASW Roper surveys usually cover only 300 people and are limited to businessmen. Some, if not all, may have been interviewed over the phone.
It should be noted that random samples allow for inferential statistics; that is, a pollster may make a projection from the data to cover the entire population. Non-scientific samples allow for only descriptive statistics; that is, they only describe the opinion or behavior of the survey group.
Rule 5: Margins of error. Survey results have a margin of error. For SWS and Pulse Asia, it is usually plus/minus 3 percent. This means that any result within this margin is statistically a tie and too close to call. Concretely, if Candidate A has 20 percent of the vote and Candidate B has 18 percent, the race is too close to call and could go either way. Media rarely makes this distinction.
Rule 6: Make sure the survey is real. Political dirty-tricks campaign operations have been known to release fabricated survey results under the names of legitimate pollsters. Since media doesn’t verify all reports, it’s best not to believe news items until you check pollsters’ websites. Both SWS and Pulse Asia have been victimized in this way in the last two months.
Rule 7: What was the question again? The press is usually quick to give you the results but oftentimes fails to even give you the question. The relevant question is “Who would you vote for?” rather than “Who do you think will win?”
Different people read surveys differently. Political parties check them (and commission them) regularly because they want to know who to field. In other words, they want to check for so-called “winnability”. Others read them to find out who to support or bankroll in an election. And still others want to simply find out who’s leading the race today. What people should not do is to read a survey so they’ll make up their minds about who to vote for. That’s not opinion measurement.
By the way, ever met anyone who’s been surveyed before?
LinkWithin
