Partisan criticism of a survey

TO THE critics dismayed by the outcome of the SWS Nov. 26-28 survey putting Davao City Mayor Rodrigo Duterte ahead of four other candidates in the presidential election race, the ultimate rejoinder is: What are your contradictory findings from other independent scientific surveys? The best way to dispute data is to demonstrate different findings from alternative sources.

Let us tell the story from the beginning. A special national survey on voting preferences that was commissioned to Social Weather Stations by William J. Lima was reported in the newspapers last Monday, Dec. 7, on the basis of certain materials released directly to the media by the sponsor the day before.

An SWS survey sponsor has the right to make public any part of his or her survey report at any time, without having to coordinate the release with SWS. SWS may anticipate that a release will happen, but cannot know exactly when it will be, unless informed by the sponsor beforehand. Almost always, the first time that SWS learns about a release is upon receiving phoned requests from the media to confirm a certain survey; that is what happened last Sunday.

SWS’ policy is not to make any statement on commissioned work to the media until after contacting the sponsor. (What if the release is a fake? It has happened before.) By late Sunday night, the sponsor’s attorney-in-fact e-mailed SWS a set of PowerPoint slides authorized for release.

The set of slides had been selected by the sponsor from among the slides in the SWS report submitted to him. SWS did not intervene in the selection. No slide in the selection had been edited. On Monday morning, the notice, “SWS confirms national survey for Mr. William J. Lima,” was posted on our website, together with the authorized slides. In the notice are the sponsor’s name, address—the contract states Pasay City, not Davao City—and mobile phone number.

The sponsor’s main interest was in learning how the entry of Mayor Duterte would affect the positions of candidates in a five-way race together with Jejomar Binay, Grace Poe, Mar Roxas, and Miriam Santiago. It is implicit in the released slide on survey question Q5: “Let us use another list that includes Rodrigo Duterte as a substitute candidate for president. Among the names in this list, who would you probably vote for as president, if the election were held now?” It is followed by “(Show List 2),” which is an instruction to the interviewer.

The mention of Duterte at the start of this item is a mere transition from a preceding question that involved, obviously, a list of candidates without Duterte. In my opinion, the critics who claim that it incorporates a “bias” are regarding Filipino voters as children whose choice for president can be changed by any newly mentioned name. Can they demonstrate that using any two lists, with their own preferred candidate entering only the second list, will surely propel their candidate into survey leadership?

The survey findings from its List 1 were not among those released by the sponsor. Any inquiries about it should be directed to him. SWS does not act as a sponsor’s agent for release of findings. It does not question the content or timing of any release. Its role is to confirm, and deny when necessary, the authenticity and correctness of any release.

If any unreleased portion of its survey would modify the tenor of its general findings, then SWS has a duty to inform the public about it. But I can declare, without reservation, that nothing in what Mr. Lima has kept private would change Mayor Duterte’s survey standing among the selected contenders at the time of the survey.

Because of the radical change in the race shown by the survey, we reviewed, in particular, the implementation of the sampling system, and did not find anything questionable in it. Yes, the change was radical; but radical changes have happened before, such as the “black swan” emergence of Noynoy Aquino in late 2009, and the steady rise of Jejomar Binay from third to first in the last vice-presidential race.

For surveys to explain changes in a race, they should specifically incorporate items that might matter, so that statistical connections can be studied. What do voters really know about the good (and bad) points of each candidate? What campaign messages are reaching them (and what are not)? What events impress them (and what events do not)?

To me, the main discovery from the Nov. 26-28 survey is the volatility of Filipino voters’ preferences for president. Perhaps the preferences will change again. Perhaps they will even change several times. Voters cannot be taken for granted. Let us see what happens in future surveys, especially the SWS own-account surveys, meant for publication as a public service.

For SWS, all survey work is nonpartisan, even when the subject matter is of great interest to political partisans. We do not expect any surveys sponsored by political partisans to be published within the agreed three-year embargo period. For sponsors to release the surveys with news good for them is only natural. For them to suppress the ones with bad news is equally natural.

We invite researchers to discover from our archives the many election-related surveys with unpleasant findings that never got published—but that may now be published, without need for anyone’s consent, since the embargoes have expired. Many survey sponsors have been thankful for obtaining scientific data relevant to their decisions of when, and when not, to join an election race.

* * *

Contact mahar.mangahas@sws.org.ph

Read more...