Sunday, February 14, 2010

Spinning Survey Results

Dr. Prospero E. de Vera

Last December 14, 2009 I reacted strongly to a Philippine Star banner story that read - Noynoy's lead in opinion polls drops, Gibo's rating surges - The Center - which claimed that there was a dramatic increase in voter support for Gibo Teodoro due to "the strength of his bold decisions on the Maguindanao massacre and his impressive showing in two presidential debates".

The newspaper headline was based on a survey made by a certain Issues and Advocacy Center headed by Ed Malay which showed that Aquino has a voter support of only 30 percent, Villar has 23 percent, and Teodoro 10 percent.

Malay was quoted as saying that the debates had harmed Noynoy because (these) "exposed the weakness and lack of depth on the basic issues by the neophyte senator when ranged against a more knowledgeable and experienced candidate such as Teodoro who topped the 1985 bar exams.” He also said that there was a detectable "stationary dive" in Aquino ratings and predicted a further tightening of the presidential race in 2010.

My reaction was carried by Philippine Star (http://www.philstar.com/Article.aspx?articleId=533342) and Manila Standard (http://www.manilastandardtoday.com/insideNews.htm?f=2009/december/17/news3.isx&d=/2009/december/17) on December 17, 2009. I criticized the disturbing inconsistency of this survey result with those of SWS and Pulse Asia, questioned the use of terms like "stationary drive" and "pro-rated", and argued that polling groups should be required to present the methods and findings of the surveys to the public.

I also asserted that it is time "that we demand more accountability from the groups behind these surveys, including their owners, directors and analysts, and who pays for their surveys". And warned that if this is not done, "we will be at the mercy of spin masters and manufactured information”.

The survey firm never rebutted my criticisms so I let it pass.

The Pulse Asia and SWS surveys of January 2010, of course, showed Gibo Teodoro with only a 5% voter support.

Yesterday, the newspapers were again abuzz with the news that Gibo Teodoro had dramatically increased his survey result from 5% to 11% in a survey conducted by StratPOLLS, a company identified in the Manila Times, Malaya and GMANews as owned and financed by former Ambassador Antonio Cabangon Chua.

StratPOLLS claims that Teodoro gained 6 percent from its September 2009 survey. Zambales Rep. Ma. Milagros Magsaysay, spokesperson of Teodoro, immediately said that the improved rating “shows that the people are beginning to appreciate the qualifications and capabilities of Gibo to be president.”

Several experts led by UP Prof. Pernia have already criticized StratPOLLS' findings for its sampling frame and the practice of averaging their data with those of SWS and PULSE even if the data gathering methods and questions are incomparable.

I've tried searching "Stratpolls" in the world wide web and can't find a web page!! And I can't electronically access the Stratpols "September 2009 survey results" either!!

Hmmm.... this smells, feels and looks like an administration PR job again.

The emergence of surveys groups who don't even have a web page, and who refuse to disclose their owners, officers, analysts, and survey data have created serious questions on the reliability of surveys in the Philippines.

And media organizations are not helping ensure transparency and accountability for these new survey upstarts because their "survey results" are legitimized by the tri-media through their news coverage.

It is time that the media, COMELEC, academic institutions, and serious civil society organizations demand transparency and accountability in polling. Survey firms must disclose information and their survey data must also be accessible to the public and not just the COMELEC and political parties are required by RA 9006.

The media can do this by only covering polling groups that make public presentations of their survey results where academicians, statisticians, and the public can question their methodology, sampling design, framing of questions, and data analysis. Those who hide behind slick media releases and media spins of political parties and supporters should be criticized and ignored.

If we cant do this, maybe its better to simply ban all polling during the election period.