Public opinion polls are playing a growing role in politics but how accurate are the results, asks Reem Leila The post 25 January Revolution economy may be flat-lining but there is at least one growth area. Following the collapse of decades of single party rule opinion polling has taken off in Egypt, though the accuracy of the findings of the sudden plethora of polls is open to question. Adel Abdel-Ghaffar, a professor at Cairo University's Mass Communication Faculty, stresses that few of the organisations currently conducting polls produce credible results. Among those that do apply a convincing methodology are Cairo University, Al-Ahram Centre for Political and Strategic Studies (ACPSS) and the National Centre for Sociological and Criminological Research (NCSCR). The increase in the number of public opinion polls after the revolution is, according to Abdel-Ghaffar, normal. "Egypt was suffering from a dictatorial regime, and during such times every effort is made to make public opinion in line with that of the ruler. This was taking place by allowing only one point of view to be heard. No true freedom of public opinion was permitted," Abdel-Ghaffar said. A recent poll conducted by ACPSS revealed that presidential candidate Ahmed Shafik, who will run despite the recent death of his wife, will benefit by 40 per cent from the banning of the three top candidates from the elections. Another candidate, Amr Moussa, will collect 27.4 per cent of voters. The poll also indicated that the Muslim Brotherhood's substitute nominee, Mohamed Mursi, is not liked by the followers of the more recogniseable MB member who was disqualified, Khairat El-Shater. And only 1.7 per cent said they would pick moderate Abdel-Moneim Abul-Fotouh. Results, says Abdel-Ghaffar, are easily skewed by asking leading questions, and given the enormous cost involved in conducting polls it is necessary, he warns, to look at who is financing the poll since they can be used to promote specific agendas and direct public policy debates. The size of the sample employed by pollsters is no guide to the accuracy of results. A large, carelessly chosen sample will be less representative of public opinion than a carefully selected smaller group of respondents, says Ahmed Nagui Qamha, head of polling at ACPSS. Most nationwide surveys are based on 2,000 to 5,000 respondents. Individual respondents must be carefully selected to replicate, as closely as possible, their weight in wider society. They are then interviewed, or answer a series of written questions designed to elicit their opinions. ACPSS's polling department has been conducting surveys of public opinion since it was created in 1997. Its polls during the 2000 and 2005 People's Assembly elections, says Qamha, displayed just a three per cent margin of error. In the 2006 presidential election, Egypt's first contested presidential poll, they were banned from conducting any surveys. During the 2011 People's Assembly elections, claims Qamha, "our findings were 98 per cent accurate. Results of our poll revealed the domination of Islamists in the parliament. Unfortunately we did not publish the results of this poll because they were scary." According to Qamha, his department was so unnerved by results that showed Islamists sweeping the People's Assembly that they did not publish the results. Qamha pointed out that public opinion polls could allow politicians to understand the public more. They could help political candidates in becoming acquainted with the public and their views about candidates "in order to correct something wrong he might be doing or if he is on the right path. Accordingly, the interviews are very important." ACPSS, says Qamha, goes to great lengths to ensure its sample groups are representative, relying heavily on data provided by the Central Agency for Public Mobilisation and Statistics (CAPMAS). "Samples must reflect reality. For instance, the population comprises 48 per cent men and 52 per cent women. Our samples must replicate this proportion, as they must replicate the actual demographic across a range of economic, social and cultural groupings." "Careless wording of questions and poorly trained interviewers can easily spoil a poll no matter how scientifically the sample has been gathered," says Qamha. Questions must be clear, so that people do not misunderstand them, and neutral, since any question which implies a specific answer can lead to a serious error in results. Qamha and Abdel-Ghaffar agree that face to face polls are the most suitable for Egypt. Phone polls, which are used widely in the US and Europe, would not work in Egypt, says Abdel-Ghaffar, where unlike Europe and North America pollsters have no access to the social and economic details of phone number holders. Internet polls are equally unreliable. Their results cannot even be said to reflect the opinions of Internet users, already a limited demographic in Egypt, since they reflect the opinions of only those who choose to fill in an online questionnaire. And as Qamha points out, there is always a danger of a single Internet user filling in multiple forms under different user names."This is what we call militias of Internet polling. They could log in and vote for a particular issue, thus providing a fake reflection of reality," added Qamha.