Counting Internet Users and calculating divides

Posted on September 22, 2010  /  2 Comments

The ITU dataset is the mother lode, mined by all. But sometimes, it is good to interrogate the quality of what the ITU produces. The most recent instance of ITU data being subject to sophisticated analysis without any attention being paid to the quality of the data is by noted ICT4D scholar, Richard Heeks.

In a previous essay, Heeks interrogated the numbers emanating from the ITU on “mobile subscriptions.” It is a pity the same was not done in the recent piece on Internet and broadband.

For example, the ITU reports that Afghanistan had 2,000 Internet subscriptions and 1,000,000 Internet users, indicating the use of a multiplier of 500. In other words, the Afghan administration is asking us to believe that each Internet connection is used by 500 people, in addition to asking us to accept nice round numbers on the subscriptions indicator.

This illustrates the biggest weakness of the ITU’s definition of an Internet User: each national administration is allowed to use a multiplier of its choice to derive the number of Internet users from the number of Internet subscribers, in the absence of demand-side surveys, the first-best way of obtaining the indicator. No low-income countries have reported demand-side survey results. Therefore, the Internet user numbers reported by the ITU are tainted by the use of arbitrary multipliers such as the 500 used by Afghanistan (this is the most outrageous multiplier we found; most are more reasonable). But the point is that it is wrong to permit national administrations which may have incentives to look good in terms of Internet connectivity to use multipliers without any rational basis. LIRNEasia is in the process of developing a practical solution to the problem of the multiplier that will be published shortly.

The Internet User number is further flawed by definitional and reporting weaknesses in the base indicator of Internet Subscriptions.

Total Internet users = Multiplier * Total Internet subscriptions
Total Internet subscriptions =Total fixed subscriptions + Total wireless broadband subscriptions

According to the recently revised World Telecommunication/ICT Indicators Definitions (2010) the definitions of these terms are below.
Total fixed (wired) Internet subscriptions = total number of Internet subscriptions with fixed (wired) Internet access, which includes dial-up and total fixed (wired) broadband subscriptions.
Total fixed (wired) broadband subscriptions = total number of subscriptions with high-speed access to the public Internet (a TCP/IP connection), at downstream speeds equal to, or greater than, 256 kbit/s.
Total wireless broadband subscriptions = sum of satellite, terrestrial fixed wireless and terrestrial mobile wireless subscriptions.

Surprisingly, even the recently improved ITU definition does not include prepaid mobile broadband subscriptions. Considering the growing number of prepaid mobile data subscriptions, especially in developing countries, this omission alone will yield significantly lower numbers of Internet subscriptions and thereby, total Internet users.

Preliminary investigations showed that in some countries such as the Maldives, even postpaid mobile data connections are not reported by operators to the administrations and therefore do not reach the ITU. In Sri Lanka, all the SIMs provided by a major operator are data-enabled. Therefore, even without a specific data plan, any customer with a data compatible mobile phone can use the Internet. These ad hoc users are not counted as Internet users by mobile operators. Therefore, the number of mobile subscriptions is underreported. As a result, the overall Internet user number is also lower than it should be.

It is necessary, before engaging is sophisticated manipulations of data to assess the quality of the data. The above discussion indicates that the current Internet User figures published by the ITU are seriously flawed because of problems in the formula, the definitions and reporting practices. Not all the problems can be resolved immediately, but they should at least be noted.


  1. Hi
    Interesting this becaus ei have always battled with those numbers as well. Do you have any odea if this / or part thereof has been resolved ?