Why researchers differ on the number of internet users?

Ajit Balakrishnan, chairman & chief executive officer of India’s leading internet site Rediff.com, has talked about why researchers differ widely in their estimates on the internet users.

US-based research and consulting firm eTForecasts, recently suggested that the number of internet users in India stands at 50.6 million. But, Internet and Mobile Association of India puts the figure at 38.5 million. Then, there is JuxtConsult which says that 27 million users in India use the internet. Even that figure may be way above the mark, if we consider the data provided by Indicus Analytics according to which the number of internet users in the country is 17 million.

AlooTechie, obviously, would not like to pick and choose any of the above mentioned data regarding the number of internet users in the country. That decision we would leave to you, our informed reader.

Still, we can start a discussion thread on why researchers differ widely in their estimates on the internet users. In that initiative, we are presenting here the edited version of an email sent to us by Ajit Balakrishnan, chairman & chief executive officer of India’s leading internet site Rediff.com.

According to Balakrishnan, the first reason is the definition of a 'user'. One definition is someone who has used some internet related service at least once in the 'relevant period'. Thus the 'user' needs not mean 'subscriber' - i.e., someone who has paid for an internet subscription. Many Indian government estimates (for example, TRAI and Ministry of IT estimates and targets) count subscribers and not users and hence come up with substantially lower numbers because for each subscriber there may be 2-100 users: 2 if it is a home and 100 if the 'subscriber' is an office or internet cafe. Government uses this subscriber measure because their source is telecom operators who can count easily only this.

A second reason, suggested by Balakrishnan for the differences, is the definition of 'relevant period'. In some studies it means 30 days, in others 45 days and in some others 'ever accessed'. Obviously the 'ever accessed' figure will be many times than that of 30 days and the 45-day access figure could be (depending on consumer behaviour and country) 50-100 per cent more than the 30-day one.

A third reason is the method used to collect the data: some studies take a random sample of census homes and go out and interview them. Others put a piece of code on a sample of user PCs and measure activity. In each of these there are challenges of sample representativeness and projection and they are not all the same across countries. India has sizeable numbers of users using the same PC (at cybercafé or office or home). Thus, the second method -- putting code on users’ PC -- involves making an assumption about the mix of sources of access and you can come up with two-times to three-times of differences depending on the assumption you make on this factor.

Finally, the number, according to the Rediff CEO, depends on whether you are counting only PC-based access or whether you count cellphones as well. Again, within cellphone-based access whether you count only GPRS/Edge user or you include SMS users as well. This is not as easy a decision as it looks because many services one associates with the internet (for example the Rediff Search for ring-tones or Exam results or email all available on SMS) are currently delivered through SMS. (Editor: What Ajit might be suggesting here is that when a user sends an SMS to a service provider like Rediff to get his exam results the user is not directly using the internet but before he gets the result in an SMS reply, someone at the service provider end must have accessed the internet to get the result from the relevant examination board’s website.) Inclusion or non-inclusion of mobile either as GPRS or as SMS can swing numbers widely.

Balakrishnan adds, “All estimates can be right depending on the above factors and on the purpose you are putting it to.”