Monday, May 14, 2007

Strange Things Do Happen

Strange Things Do Happen: In 2004, the SWS/ABSCBN Day of Election Exit Poll correctly predicted eleven of the twelve eventual official winners of the Senatorial Race from Mar Roxas at No. 1 to Rodolfo Biazon at No. 12 (by only 6,000 votes over Robert Barbers at heartbreak 13). However, the SWS Exit poll completely missed the 8th place Senate finish of Alfredo Lim, whom the exit poll placed at 14th-15th place. This election year, the SWS/PDI prediction is for a 6-4-2 GO-TU-Ind finish. Getting all 12 winners right would certainly bespeak of "world class" public opinion polling--the best that could possibly be achieved. In a sense, it is not only GO or TU gunning for that mythical 12-0. But SWS has a better chance than either of achieving it!

There is certainly an admirable kind of CHUTZPAH involved in being a Public Opinion Pollster. Making brazen predictions about who will win Senatorial elections and by how much, at the risk of gaining a reputation for quackery, requires that the pollster have supreme confidence in the science of statistics that he is employing. Surveys may seem indistinguishable from the magic to the masses, although done correctly, there is no need for tricks in nailing the results of an election with an exit poll or pre-election survey, and endless prestige. Public affirmation of a pollster's credibility and reliability seems to be the number of privately commissioned surveys that follow from insecure ambitions anxious to avoid penury through vainglorious delusions about their chances of electoral victory.

DR. MAHAR MANGAHAS of the Social Weather Stations made the perhaps little-noticed remark on ABSCBN News (ANC) last week that "in the last three years" SWS has "become independent" of its data collector, Trends TNS--the little known market research firm that used to collect the raw survey data for the SWS periodic monthly and quarterly surveys. SWS has apparently decided to develop its own organizational capability to do the field work previously outsourced to Trends like selecting the random sample of respondents, distributing and explaining the survey questionnaires, interviewing the respondents, organizing and tallying the results, etc. I welcome this capability expansion on the part of SWS, as it will lead to potentially more security and integrity for their future polls. (It also seems to be helping the SWS bottom line: as an unexpected bonus, SWS is apparently getting a lot of new business for pure market research--the main line business of Trends.).

After hearing Mahar's remark, I could not help but think the breakup of a long-standing relationship between Pollster and Surveyor must have something to do with that disastrous SWS day of election survey for President in 2004. Something went awfully wrong with that survey that they had to convene a committee to figure out what.

Now that SWS is capable of doing its own survey field work and raw data collection, they can secure their raw data better and ensure its integrity from the source. When SWS relied on a different organization to collect this data, I believe they were vulnerable to a knowledgeable attack or special operation in which survey data could be corrupted for some intended effect.

4 comments:

PJ Punla said...

aah, I remember the days when I was taking the relevant courses in Communication Research at UP MassComm.

how cynical I was then about the possibility of data corruption and manipulation.

how cynical I am now of the election and its counts, as it's a nearly foregone conclusion that those who win will be accused of cheating, and those who lose will accuse their rivals of cheating.

Amadeo said...

Presidential exit polling? 2004?

Shades of the US presidential election of 2004, when early exit polling showed Kerry in a comfortable lead against Bush.

When the actual count started showing a Bush lead, charges of cheating and/or fabrication came just as quick.

Marcus Aurelius said...

DJB,

You should reprise the little gadonkin experiment you posted a year or so ago on determining the color population of a mixed set of color balls in a swimming pool.

It was helpful in my own understanding of polls as it allows one to understand the basic concepts of polling most notably sampling.

lateralus said...

Yup, ranking Lim at 14th/15th was horrendous for the polling body (SWS). But I have a feeling that Cayetano will get no higher than the 13th spot. hehe