“Fast is fine, but accuracy is everything.” ― Xenophon (Greek historian, author of the Anabasis, BC 431-350)
Internet penetration rates are often used to structure ICT policy and to attract investment. If internet user data is important to economic growth, why isn’t it more accurate?
As most are aware, there is no sure-fire way to quantify the number of Internet users in countries where household Internet access is low. The reason being multiple users per subscription or access point (often a café with dozens of users). The same challenge goes toward counting mobile subscriptions, since many mobile owners rely on multiple carriers.
The ITU puts some effort into estimating the number of Internet users, but the organization’s data lags by two years. Still, it has become standard fare in reports, whitepapers, benchmarks, and infographics. The ITU realizes the best method for data collection is the old fashioned questionnaire, but what happens when that method is rejected at the local level? Thus is the other challenge we face with finding “good” data.
On top of timeliness, there is no true “apples to apples” method for comparing Internet penetration rates across nations. Issues arise with how the ITU stats are sourced. The ITU relies upon government websites and operators’ annual reports for nations that do not respond to the annual questionnaire. It’s not reasonable to assume all operators collect data using the same means.
Any Data better than no Data?
Internet World Stats routinely publishes updated Facebook user data at a country level. This data is some of the best available – it’s direct from Facebook itself and can be determined in realt-time. Of course, the set is limited to users who report a location, and at that, their actual location. Still, one would imagine that the Facebook user data under-reports the number of real Facebook users in a given nation since many users do not provide more than just a name and photo. However, we find the opposite if we are to merge current Facebook and ITU data.
In October 2010, we, oAfrica, observed the discrepancy, noting how “Botswana and Sierra Leone have surprisingly high Facebook usage rates that are in the vicinity of South Africa’s 60%.” Over one year later, Botswana and Sierra Leone have Facebook usage rates of over 120% – meaning that there are more Facebook users than Internet users. This is logically impossible. In fact, if we are to hold the IWS data at face value, Botswana, Central African Republic, DRC, Equatorial Guinea, Ethiopia, Libya, Mauritania, Mauritius, Namibia, Sierra Leone all have more Facebook users than ITU data suggests.
The ITU should at least update a subset of data to reflect third-party sources such as Facebook. Even if it means sacrificing traditional methodology, another set of data could be updated throughout the year with the caveat that it is sourced using mathematically risky methods.
National Regulator vs. ITU
Tanzania Communications Regulatory Authority (TCRA) recently cited an 11% Internet penetration rate, with 5 million Internet users in the country. At the same time, the ITU quotes fewer than 700,000 Internet users in Tanzania as of June 2010. Even in booming East Africa, Internet adoption rates haven’t hit 700% in the past year-and-a-half as suggested here. Either the ITU’s questionnaire in 2010 produced extremely conservative findings or Tanzania’s regulator needs to re-evaluate its subscriber data. The ITU must have conferred with TCRA when they last gathered data, but the discrepancysuggests otherwise.
ITU Internet numbers are rightfully authoritative and grab the attention of the public, but they should be used with care. Instead, consider citing local data when possible. It’s more targeted, more pertinent, and usually more interesting. Regional habits vary and maynot be applicable for the nation at-large. What’s true for Nairobi is not true for Mombasa, let alone rural Kenya. National data would most likely combine all of the data into one bucket, thus hiding subtle trends.
On the bright-side, regional surveys and national regulators often provide relatively accurate data (be sure to read their methodologies first). And it’s more granular than simply the number of Internet users. And it’s current. Unfortunately, such data is difficult to find since the resources needed to conduct a survey are scare. For now, rough estimates will suffice for creating policy, attracting investment, and charting growth, but the time will come when precise data is necessary. After all, stakeholders with the most accurate pulse on ICT growth will have a leg up on the competition.
Cover Image Credit: Alan Davey