Friend-of-the-blog Gabriel Koehler-Derrick does some really neat stuff with Google to track prominent personalities in religious currents, politics, and society in the Middle East. In this commentary he sent us, Gabriel looks at Google as an alternative indication of the popularity (or interest in) the various candidates in the Egyptian presidential elections. A PDF version of this article, which includes graphs that are tricky to transpose to the web, is here (275kb).
With the approach of Egypt’s presidential elections on Wednesday, a variety of polls have been published trying to anticipate the outright winner, or at least identify which two candidates are capable of winning enough votes to force a runoff election. Given the challenges associated with polling in Egypt, the historic nature of the election, and a confusing series of legal rulings that have dramatically shaken up the field of contestants, it is not surprising that the outcome remains unclear. While far from perfect, data from internet search trends suggest a far less ambiguous outcome: Amr Moussa is comfortably in the lead and Muhammad Morsi is the candidate most likely to face him should there be a runoff.
Anyone familiar with the telecommunications industry in Egypt might question the utility of using data derived from internet searches to better understand political developments. While internet penetration rates have grown impressively, according to a recent survey conducted by A.C. Nielsen for Google’s MENA office, only about 39% of Egyptians have “regular access” (defined in the survey as logging in once a month) to the internet. The data from the A.C. Nielsen survey also show that Egypt’s community of internet users are disproportionately male, and younger than average, with the 15 to 24 and 25 to 34 age cohorts being particularly well represented. To the best of my knowledge, credible statistics about the income and education level of Egypt’s internet users are not publically available, but it doesn’t seem too much of a stretch to presume that typical internet users are skewed towards urban areas, and better educated and significantly wealthier than national averages. Because of these challenges, data derived from internet searches cannot be considered statistically representative of the Egyptian population.
Despite these drawbacks, internet search data enjoys a number of advantages for examining the presidential race. First, the number of data points for any time period is huge. A back of the envelope calculation, based on the Nielsen survey and some basic population data from the UN, suggests that in Egypt, Google gets almost 26 million searches a day. While only a tiny fraction of these searches are politically related, nine out of the “top 10 rising people” in Google’s 2011 Zeitgeist survey of Egypt’s search trends were connected to the revolution or politics more broadly, indicating just how influential political developments in 2011 were on search trends in Egypt. By way of comparison, none of the “top 10 rising people searches” in Turkey has anything to do with politics, and only one of the “top 10 rising people searches” in Canada , former leader of the National Democratic Party (NDP) Jack Layton. Data from Google AdWords, provides an updated 30-day average of the number of searches for a given term, shows some impressive averages for each of the top presidential contenders. This is crucial because it provides a sense of scale for the Insights for Search data cited below, which uses normalized results not raw numbers to plot the trend lines for the various candidates.