LIRNEasia is a regional ICT policy and regulation think tank active across the Asia Pacific

Call for proposals: CPRsouth 2017. For more details see announcement.

Work with us: We are currently recruiting for the position of Big Data Researcher.

How many Internet users?

As far we knew, this number is collected on the basis of demand-side surveys. When such surveys were not available, “administrations” (Ministries or Regulatory Agencies tasked with the job) would submit estimates, based the number of subscriber and a multiplier. We’ve spent untold hours on the phone, trying to wheedle these numbers out. We even published a peer-reviewed article proposing an alternative (and, in our opinion, superior) method.

But little did we (and the peer reviewers, and those collating the data at the ITU) know. There was a superb data source that reported data on a second-by-second basis. Not like the ITU where the data was updated once a year, this changes before one’s eyes. I couldn’t tell you how many Internet users there are in the world, because the number changes from the time I write to when I hit publish. That is intuitively true, but how did they pull it off? There is no methodology note. They simply say they take data from the usual sources, ITU, World Bank, etc.

So this caused me to bring out the old nose test. According to the site, Mali has a Internet user percentage of 75.23. That is, out every 100 Malians, 75 use the Internet on a regular basis. This is the country that gave the world ITU Secretary General Hamadoun Toure, after all.

But still, my nose was twitching. Mali’s GNI per capita (Atlas method, current USD) is USD 660. It’s a low-income country. Its poverty headcount has decreased, but in 2010 it stood at 43.6 percent. It had minus economic growth in 2012. I am using data from a source listed by the Internetstatslive folk: the World Bank.

How could a country in that kind of shape (and not fully in control of its territory because of the Touareg rebels) achieve this miraculous Internet penetration, zooming ahead of the usual leaders in Africa (Mauritius, at a measly 6.14% in 2014 and South Africa, at a much better 46.88%)? It had not only overtaken the African powerhouses, it had even surpassed Spain,at 74.38%). It had a growth rate of 16 percent last year. My nose was, by now, figuratively aflame.

But does one mistake a whole data source destroy? Sadly, yes.

But since so many innocents were relying on the slick presentation (“if it’s on the Internet it must be true; if the design is good and they claim to have been cited by the Web Foundation, it must be doubly true”), I probed a little further, now looking at Sri Lanka.

I couldn’t assess their Internet numbers, which were below my estimate of Internet users in Sri Lanka and above what the TRC reports for Internet subscriptions. But I saw a red flag in the population numbers. This is kind of important because what matters for most people is not the absolute number of Internet users, but that number corrected for population, i.e., divided by population and multiplied by 100. The population was overstated by over 1 million. That means that the Internet Users/100 number is understated.

The fraudulent website claims that Sri Lanka has 21.45 million people. The most reliable information on population comes from the 2011-12 Census which gives the population as 20.27 million people in 2012. Could we have added a million plus in one year? Unlikely given our fertility rates, but let’s check from another authoritative source. What does the World Bank, their stated source of population information say? They say 20.33 million for 2012.

Oh, and by the way, the Lanka census reports Internet use at household level (which means that Internet user numbers must be higher).

Access to the Internet from within the home is available to 11.4% of households. 12.7 percent households report Internet use from outside the home. That means that a reliable demand-side survey using random sampling gives a household penetration level of 24.1%, which is higher than what these charlatans report at the individual level: 19.9%. If anything, my estimate of 25 percent Internet users is too low.

So what is the conclusion? If it looks too good to be true, it is. I wouldn’t trust one number in this website.


Research Mailing List

Enter your email for research updates:


Flickr Photos