A good time for local opinion polling

The next nine months, up to the elections in May 2016, is a very good time for private—i.e., nongovernmental—universities and colleges with basic social science capability to engage in election-pertinent opinion polling, in their respective geographical areas, for the sake of public information. I address this piece to private institutions, since they are not as subject to political pressure as those in the public sector that also have the technical capability.

People like opinion polls. SWS surveys have found that Filipinos find opinion polls, on whatever topic, quite interesting, and would like their political leaders to heed them. Filipinos believe that opinion polls are generally accurate.

The thing that still amazes most Filipinos—as well as most Americans, by the way—is how a sample of only 1,000 respondents is often enough to predict the result of a national election.

Politicians are very interested in opinion polls. Aspirants for elected office, whether they admit it or not, are extremely interested in opinion polls. They savor findings favorable to them. They carefully scan the people’s attitudes for potential opportunities.

When politicians cast aspersions on unfavorable findings, it is often political posturing. Some say they don’t believe surveys. Yet they commission them (or get friends to commission them) for their private use. Politicians are repeat customers, even when initial survey findings are discouraging. They don’t want to lose elections because their opponents were more scientific.

In all countries, citizens in general, as well as politicians in particular, use success in predicting elections as the litmus test of survey quality. Local academic institutions will boost their research reputations immensely by using the election season to take this test, and then passing it.

The technology of polling is straightforward. Of course, it is essential that the respondents of a survey be drawn at random. Random selection is a statistical principle. Opinions of self-selected respondents are a waste of time to read.

The sample size of a survey is like the number of pixels available in a photograph. The more pixels, the sharper the picture; the fewer pixels, the grainier. A sample of 1,200 (the standard SWS national sample) gives percentages accurate up to plus/minus 3 points. A sample of 300 (the SWS sample for a subnational area like Mindanao) has errors of plus/minus 6 points. But reliable district, provincial, or city close-ups cannot be obtained from a national survey. Such close-ups require dedicated local surveys, which are jobs ideal for local academic institutions.

Actually, local election races are often lopsided and easy to forecast. So local researchers should not be timid. The 95-percent significance level of the above error margins implies that 19 out of 20 surveys using them will arrive at the same forecast. A local election survey does not need a perfect score on the many races (district representative, governor, vice governor, board members, mayor, vice mayor, councilors) that it considers in order to be a success.

Exit polling and election surveying are protected by law. Local researchers should know of the Supreme Court ruling (ABS-CBN versus Comelec, GR No. 133486, Jan. 8, 2000) that exit polls are entitled to constitutional protection:

“The holding of exit polls and the dissemination of their results through mass media constitute an essential part of the freedoms of speech and of the press. Hence, the Comelec cannot ban them totally in the guise of promoting clean, honest, orderly and credible elections. Quite the contrary, exit polls—properly conducted and publicized—can be vital tools in eliminating the evils of election-fixing and fraud.”

They should also know of the Supreme Court nullification (SWS versus Comelec, GR No. 147571, May 5, 2001) of the Fair Election Act’s ban on publication of surveys before elections, for violating the right of freedom of expression:

“To sustain the ban on survey results would sanction the censorship of all speaking by candidates in an election on the ground that the usual bombasts and hyperbolic claims made during the campaign can confuse the voters and thus debase the electoral process.”

Surveys can be revenue-generating. Finally, local academic institutions should appreciate that doing their own survey research could be a financially self-sustaining activity. For one thing, it is advisable to plan a survey as a multitopic affair, since it is a powerful tool for obtaining data on many matters, some of which would have market value for potential local sponsors.

It may be possible for the entire cost of a survey to be defrayed by such sponsors, even if many questionnaire items—i.e., those election-related and meant for publication—are unsponsored. A contract to keep sponsored items confidential or solely for the use of their sponsors would not be unethical. After all, if the said items had not been sponsored, their data would not have been collected in the first place.

In the United States, there are very many universities with research units that support themselves solely on contracts to do scientific surveys for agencies of various levels of government, as well as for private businesses. In all countries, including the Philippines, the growing appreciation among decision-makers that their policies and programs need to be evidence-based will help survey research to pay for itself.

* * *

Contact mahar.mangahas@sws.org.ph.

Read more...