By Wasiq Ali
This article appeared in The News on August 20, 2009
Opinion polls are new to Pakistan but they have been around for decades in the West. Even then, they are seen for what they are – a snapshot in time of how people feel. Opinion polls are not a substitute for elections and no one in any developed democracy demands that the result of an election be reversed simply because of opinion polls. Unfortunately, some elements full of hatred against the current elected government and particularly President Asif Zardari, are trying to do just that based on recent opinion polls. It is important, therefore, to discuss polling and its relationship with democracy in Pakistan.
Approval ratings for British Prime Minister Gordon Brown, according to opinion polls, is down to around 18 percent. But there is no clamour for a minus-one formula in Britain. Everyone understands that Prime Minister Brown was elected for a specific term and unless the majority of his party’s members of parliament change their party leader he will continue to serve as prime minister until the next election.
Similarly, US President Barack Obama’s approval ratings have dipped since he swept into office in November 2008. Obama will probably use the polls to figure out how to improve communication with the American people. The polls will not, however, affect his position as president until the next presidential election in 2012.
In India, opinion polls before general elections early this year predicted a hung parliament. The election results, however, put in office a strong Congress-led government, proving the limitation of opinion polling.
Much is being made of the recent polls in Pakistan, which show negative ratings for President Asif Ali Zardari and the Pakistan Peoples Party (PPP)-led government. Before we get swayed away by the poll numbers we should examine the problems with opinion polling in Pakistan as well.
Opinion polling was originally designed to represent the views of a population by choosing a sample and asking them a series of questions. The sample was to be selected in a way that it represented, to the extent possible, the entire population. This would help extrapolating generalities from the sample to the population at large. Thus, a good polling sample would account for the ratio of different ethnicities, professions, religious and sectarian affiliation and gender.
Pakistan’s polling organizations admit themselves that their samples do not reflect Pakistan’s diversity adequately but this admission usually comes in a separate section of their reports and is often overlooked by the media.
Furthermore, in the West polling organisations openly acknowledge any political affiliation or sponsorship. We hear of Democratic Party pollster Mark Penn and Republican pollster Frank Luntz. The affiliation does not mean the polling data is inaccurate but it warns analysts against any possible tilt in the manner of framing of questions.
Pakistan’s oldest polling organization, Gallup Pakistan, was founded by Dr Ijaz Gilani, a competent man, whose polling reports do not frankly acknowledge his association with the Jamaat-e-Islami. The very fact that the primary field force for Gallup’s polling is recruited from Jamaat-e-Islami’s cadres could affect the outcome of the poll in the view of some. This is not in any way meant to cast aspersion on Gallup’s polling in Pakistan, just to suggest the need for full disclosure. Polls can only be accurate where the framing of the questions has been less than subjective and the public has a right to know the subjectivities of those conducting any poll.
A common mistake in Pakistani polls, which makes them less accurate, relates to provincial and urban-rural weight. For example, the latest Pew Poll acknowledges that its sample of 1,254 adults was “disproportionately urban.” The Pew Poll sample was 55% urban even though Pakistan is only 33% urban. The reason for the disproportionately urban sampling was stated to be the “greater heterogeneity of the urban population.”
In its methodology section, the Pew poll states, “In addition to sampling error one should bear in mind that questions wording and practical difficulties in conducting surveys can introduce errors or bias into the findings of opinion polls.”
The Gallup-Al Jazeera poll also had given Punjab 60% in the polling sample whereas Punjab only accounts for 52% of Pakistan’s population.
Such anomalies fail to recognize the fundamental realities of Pakistan. First, Nawaz Sharif and Pakistan Muslim League-N are primarily based in urban Punjab. Any poll giving more weight to urban Punjab would unduly favour Nawaz Sharif and his party. Secondly, urban Pakistan is more affected by the themes in the media. Since the Pakistani media has been intrinsically hostile to President Zardari that hostility is reflected in urban opinions.
Thirdly, polling is difficult in areas such as the Federally Administered Tribal Areas (FATA). For example, the limited polling done in FATA affirms that people there support drone attacks because they get rid of people who are making their lives miserable.
But in the rest of the country, especially the urban areas, there is greater hostility to drone attacks largely because of the media drumming up anti-Americanism.
A detailed study of the Pew Poll methodology also indicates that there is no accounting for proportions of well-to-do versus poor and Muslims versus non-Muslim minorities. All these factors mean that just as the Gallup and other polls before the February 2008 elections failed to take into account rural and minority opinions they are doing the same even now.
Most polls before the 2008 elections predicted fewer seats for the Pakistan Peoples Party (PPP) and a lot more for the two Pakistan Muslim League factions. But we all know that the election result was different. So if President Zardari is still smiling after unfavorable opinion poll results he probably has good reason. Public opinion can change and if Pakistan can stay on the democratic course, which it should, then much can happen to alter public opinion between now and the next election due in February 2013. It is election results and not opinion polls that matter, however much hysterical anchors and columnists might drum up the opinion poll surveys.