In order to get it first, they may end up getting it wrong. As Pakistanis head to the elections in May, pollsters and others are out in full swing trying to forecast how Pakistanis will vote. The forecasts are, however, scattered all over the place, suggesting that most, if not all, will be wrong.
The amateurs and professionals have all jumped into the fray. Gallup’s Pakistan-based office is showing the Pakistan Muslim League-Nawaz to be in a comfortable lead enjoying the support of 41 per cent of the eligible voters. On the other hand is the Sustainable Development Policy Institute (SDPI), a respected Pakistan-based NGO, that has put Pakistan Peoples Party in an uncomfortable lead with the support of 29 per cent of the eligible voters.
At stake are 272 seats in the National Assembly and another 728 seats in the provincial legislatures. Between now and the eve of the elections, numerous new polls will be released predicting one or the other party in the lead. Most, if not all, predictions will be wrong. Not because the work done by the pollsters is shoddy, but because forecasting is not an exact science. To paraphrase the famed statistician George E.P. Box, who died recently, all forecasts are wrong, but some are useful.
The latest poll in Pakistan by Gallup-Pildat suggests that the PML-N is in a comfortable lead with 41 per cent of respondents indicating that they would vote for the PML-N. Contrast this with the SDPI poll that puts the PML-N in the second position with the support of 25 per cent of the respondents. Another poll by the International Republican Institute (IRI), a US-based think tank, also shows the PML-N leading, but with the support of only 32 per cent of the respondents.
There are several reasons for the difference in these forecasts. One could be the timing of the polls. The SDPI poll was conducted between October and December of 2012, the IRI poll was conducted in November 2012, and the Gallup-Pildat poll was conducted in February 2013. Each poll, even if it followed correct methodology, is representing the electorate’s mood at three different time periods. Since there is no reason to assume that the political fortunes are static over time, it is therefore possible that the polls may be reporting on the shifting of the popular opinion as the election date nears.
Another important factor that goes to the heart of survey research is the size of surveyed population and the representativeness of the sample. A random sample, or a variation of it, is necessary to ensure that the surveyed individuals are representative of the underlying population. Thus a survey that interviews respondents in overwhelmingly urban settings will fail to capture the electorate’s mood in rural areas. Similarly, an Internet-based survey will fail to report on the preferences of those who do not use the Internet, who are at least 85-plus per cent of the population in Pakistan.
Even when a sample is randomly selected, it has to be of an adequate size to offer some statistical validity. A very small sample will result in a very large margin of error. At the same time, a very large sample size does not necessarily offer better statistical accuracy. The SDPI poll is based on a very small, likely an unrepresentative, sample of 1,283 eligible voters, whereas the IRI poll is based on responses from 4,997 eligible voters. Given that there are approximately 85 million voters in Pakistan, and that there is tremendous heterogeneity in income, languages, culture, and religious and political persuasions, a sample size of 1,300 is likely to be non-representative of the entire nation. A sample size of 2,000 to 5,000 is more appropriate for a similar survey in Pakistan.
Another key player in determining public opinion in Pakistan is the Washington-based Pew Research Center, which has, over the years, conducted several surveys in Pakistan. In one of the questions, the Pew Center asks the respondents in Pakistan if they held a positive view of specific political and other personalities in Pakistan. They found that Imran Khan enjoyed the most favourable view amongst the Pakistanis. Between 2010 and 2012, those who held a favourable view of Imran Khan jumped from 52 per cent to 70 per cent. During the same time period, Nawaz Sharif suffered a slight decline in popularity since those who held a favourable view of him dropped from 71 per cent to 62 per cent. The most to lose in popularity was President Asif Ali Zardari who dropped from a one-time high of 64 per cent holding a favourable view of him in 2008 to a mere 14 per cent in 2012.
If the results of various polls are reviewed with care, it appears that the PPP and its leaders are facing severe challenges in winning over the electorate. At the same time, most polls suggest that the PML-N may have a lead over the rest for now.
A lot can change between today and May 11 when the citizens exercise their right to vote. At the end of the day, the opinion polls are nothing more than the opinions of a few recorded by some who may have a vested interest in the outcome of the opinion polls. What really matters is how the millions exercise their right to elect their representatives on May 11.