Methodology

Facebook data

The number of Facebook users in all 22 Arab countries, was collected periodically between June 25, 2013 and May 1, 2014, with breakdowns for men and women, and in the following age brackets— youth (15-29), and adults (30 and over).

Raw data on for all Arab countries was collected and aggregated based on Facebook’s official data (Group A), excluding Syria and Sudan (Group B), for which data was extracted from a source other than Facebook. Because of US imposed technology export sanctions, no data on the number of Facebook users in Syria and Sudan is available. The actual number of Facebook users in Syria (in November 2007) was located through online research; no such data for Sudan was found. For that reason, all Facebook data on Sudan in this report was estimated using the daily growth rate of Arab users (calculated from the Group A countries over different periods different periods for different reports - between January 5 and April 5, 2011; between January 3 and April 3, 2012; between January 7^th and February 18^th; and August 10 and November 9, 2013). This rate was used to calculate the number of users in group B, for consistency’s sake, to ensure a smaller error margin than using the growth rates of similar countries for each individual country.

For Syria, specifically, after February 7, 2011, when social media sites were no longer blocked in the country, a different daily growth rate was used to reflect the ensuing surge in growth number of Facebook users. This rate was based on the average daily growth rate in Yemen, which has a similar ICT and socio-economic indicators as Syria. In addition, a one-off factor was added to the calculation of the Syria growth rate after lifting the ban on social media website on February 7, 2011. This was estimated based on the surge in number of Facebook users in Egypt after a similar Internet ban was lifted on 2nd February 2011.

In order to maintain accuracy, the average daily growth rate for Yemen was recalculated at several points in time and applied to generate the number of Syrian Facebook users.

It should be noted that for all charts in this paper, the numbers of Facebook users in Syria and Sudan are estimates, while the numbers for remaining countries were compiled based on official Facebook data.

Twitter data

The number of Twitter users, number of tweets, and top trends in all 22 Arab countries, in addition to Iran, Israel and Turkey, was estimated in the month of March 2014 by sampling 733,000 Twitter users and 2.47 million tweets. The study was conducted using a specially developed Twitter API. For the first time, gender breakdowns of Twitter users in each country were included in the data collection. Additionally, data related to government Twitter accounts were collected during the month of March 2014 as well.

Two sampling methods were used:
1. Trend & volume data was collected by sampling 1% of the whole of Twitter traffic, and filtering for location.
2. Users were sampled by randomly inspecting user ID numbers. This allows for finding information on both active and inactive users.
The population estimates come from combining these two data sources (sample (2) gives a picture of user behavior, which helps assess the fraction of the population that was picked up in sample (1)). An estimated correction was applied for unlocatable users.

Data collection was done by filtering the Twitter sample stream (which provides 1% of all tweets as they happen) for tweets from the right country. An unbiased distribution of tweet frequency was obtained by random sampling of the user space. Geo Location (identifying the country from a location) was done by filtering tweets with location information using a mixture of Yahoo and Google’s geolocation services, plus a local database & some extra clean-up for mistakes by one of these services.

Population estimation was done by estimating the probability of seeing a given user appear in the stream, given the sampling period, tweet-frequency distribution, and the stream behavior (witnessed in the 1% sample of tweets; assumed unbiased). A second correction was applied for un-locatable users.

In this report, an additional correction was applied for unlocatable users in Saudi Arabia, incorporating the number of virtual private network (VPN) users that may have been incorrectly geo-located by passive web analytics. According to Global Web Index findings, 28% of internet users aged 16-64 globally say that they have used VPNs (virtual private networks) to go online. Some 44% of this group report using them in order to access social platforms like Facebook and Twitter – a figure which translates to nearly 185 million people[1]. Although there are other Arab countries with a significant percentage of VPN users, not all VPN users mask their location, and therefore would have been correctly accounted for. Users in certain countries, such as Saudi Arabia, have both higher incidences of masking their locations, and larger numbers of undisclosed VPN users, due to a stricter regulatory environment surrounding Internet access. This in turn would lead to an inaccurate number of VPN users, and a larger number of unlocatable users in Saudi. As such, a correction was applied to the estimated number of Twitter users, based on the top market for VPN users, which would more accurately reflect the number of users in Saudi Arabia.

LinkedIn data

The number of LinkedIn users in 12 Arab countries: Algeria, Bahrain, Egypt, Jordan, Kuwait, Lebanon, Morocco, Oman, Qatar, Saudi Arabia, Tunisia and the UAE, was collected periodically between June 25, 2013 and May 1, 2014, with breakdowns for men and women, and in the following age brackets: 18-24, 25-34, 35-54, 55 and above. Due to changes in the examined platform analyzed here, estimated numbers of users from several countries differ in previous report findings.

This could be due to changes, or corrections, in the criteria of geo-locating users by the platform itself. Attempts by the authors to examine the changes in the criteria with the platform where not successful, however, our analysis suggest that the current number in the Arab countries examined are more geographically representative.

Regional Survey on Social Media and Citizen Engagement

The regional online survey was administered in 22 Arab countries and ran from Feb – May 2014. Respondents numbered a total of 3654.
This sample was comprised of two smaller samples: the ‘no demographics’ group of respondents, who did not provide their demographic data in the survey, and the group of respondents who provided their ‘demographic’ data. They numbered at 692 respondents and 2962 respondents, respectively.
In the ‘demographics’ sample, around 25% of respondents were female while 75% were male. Around 13% of respondents were between the ages of 18 and 24, 44% were between the ages of 25 and 34, and 23% were between 35 and 44. Around 90% of respondents said they have children. The largest number of responses came from Egypt (24%), Saudi Arabia (20%), UAE (11%), Algeria (9%), Sudan (7%) and Jordan (7%).

Although the ‘no-demographics’ sample did not fill out demographics questions, their locations were identified and analyzed, and their geographical distribution mirrored that of the ‘demographics’. Gender and age statistics could not be collected for the no demographics’ sample, but given the even geographical distribution of the entire sample, one can assume this applies to gender and age as well.

For certain question breakdowns, respondents were also divided into country income brackets, as per the World Bank’s classifications: http://data.worldbank.org/about/country-classifications. These corresponded to:
·         High Income: Bahrain, Kuwait, Oman, Qatar, Saudi Arabia, UAE
·         Upper Middle Income: Algeria, Iraq, Jordan Lebanon, Libya, and Tunisia
·         Lower Middle Income: Egypt, Mauritania, Morocco, Palestine, Sudan, Syria and Yemen
·         Low income countries (Comoro, Djibouti, and Somalia) were excluded from this breakdown due to their small, and therefore unrepresentative, sample size, which would increase the margin of error.

The charts depicting the findings related to these breakdowns only included the ‘demographics’ sample, given that the respondents could not be classified into country income brackets without the demographic information. However, comparing the responses of ‘no-demographics’ sample to ‘demographics’ sample showed a high similarity and would indicate the possibility of generalizing the ‘demographics’ breakdowns to the entire sample.

Some questions specifically targeted government officials and as such were only answered by a smaller pool of respondents who had stated that they work in the public sector. The geographic distribution of this sample (708 respondents) differs from the full sample, with majority of respondents coming from: Egypt (21%), Algeria (19%), Saudi Arabia (12%), Sudan (11%), and UAE (10%).

[1] Global Web Index (2014).