For many voters, the results of surveys before the elections were shocking, reports Reem Leila Preceding the two-day presidential elections which began on 23 May and which were the first true democratic elections in Egyptian history, were several opinion polls conducted by various centres trying to predict the results. But the polls came up drastically short. In the most recent poll carried out on 20 May by the Al-Ahram Centre for Political and Strategic Studies (ACPSS), Amr Moussa was in the lead with 40.8 per cent support, while Ahmed Shafik was second with 19.9, Abdel-Moneim Abul-Fotouh 17.8 per cent, and Mohamed Mursi 9.4 per cent, followed by Hamdeen Sabahi who came fifth with seven per cent. The ACPSS had published regular face-to-face surveys for eight successive weeks in a sample ranging from between 1,200 to 2,000 respondents. In another survey conducted by the Cabinet's Information and Decision Support Centre (IDSC), Shafik was first with 12 per cent, Moussa second with 11 per cent, Abul-Fotouh third with nine per cent, Mursi only six per cent and Sabahi five per cent. The survey was based on a sample of 1,390 respondents interviewed during mid-May. Another opinion poll conducted by Al-Basira Centre headed by Maged Othman, former IDSC head, Shafik was the leading candidate followed by Moussa, Abul-Fotouh in third and Mursi fourth. The survey was held for 2,200 respondent in an interview over land lines and mobiles. In the end, Mursi came first, followed by Shafik, Sabahi, Abul-Fotouh and Moussa. The results only made Egyptians wary of the centres. Adel Abdel-Ghaffar, a professor of public opinion at the Faculty of Mass Communication in Cairo University believes the dramatic difference in poll numbers may be due to the difference in sampling methods. "There are many centres which are still learning about polling in Egypt," Abdel-Ghaffar said, adding the best method which suits Egyptian society when conducting opinion polls is face-to-face. There were several surveys conducted over the phone but he said this was not an accurate methodology for Egyptian society. "Housewives and the unemployed tend to answer the phone more in the morning. Landlines are more prevalent in the cities, and there is a higher proportion of mobile phones in rural areas. This methodology results in drawing up inaccurate samples because society is not only housewives, the unemployed and farmers," Abdel-Ghaffar argued. At the same time, according to Abdel-Ghaffar, different times of conducting the surveys would definitely produce inaccurate results. "The survey could have been done following a certain incident which might affect the result of poll. Therefore, choosing the right time to conduct a poll is essential," he added. Many experts believe the type of sample voters interviewed -- especially their socio-economic backgrounds -- can also end up distorting results and misrepresenting public opinion. Ali Leila, professor of sociology at Ain Shams University, said all pollsters should make sure to use representative samples, ones that includes all socio-economic classes, political orientations and educational strata, as well as both genders. "Opinion polls based on unrepresentative voter samples are worthless for gauging the true extent of candidates' respective popularity among the voting public. All of the samples of different centres were too small; they were unrepresentative of the entire population," Leila said. The slanted nature of the polling might have given a faulty impression about presidential candidates. Leila questioned the competence, credibility and the political hues of the centres conducting the opinion polls. "Due to the inaccurate results of opinion polls, nobody can know for sure exactly where the two candidates going into the run-off stand." He stressed the importance of posing poll questions objectively, without leading respondents to desired responses -- a tactic often employed by pollsters. They also point out that personal interviews are much more reliable than telephone surveys in this regard. "The worst way to collect data is via telephone. Respondents usually want to finish the call quickly, and are therefore more prone to simply tell the interviewer what he or she wants to hear, just to finish the call," Leila argued.
