Monday, 10 December 2007

Is the on-line research feasible in Slovenia?

I've received a couple of reactions to my post regarding Eurostat's recent on-line Internet penetration figures in Europe (I actually received two e-mails to be exact). Both reactions came from colleagues from Eastern Europe. Both of you made a similar point about the fact that it's interesting to read about Research 2.0, but your countries still have very low penetration of Internet users (your examples: Bulgaria 19% and Romania 22%) so it seems little relevant as a method. These comments triggered a question: Is on-line research by default impossible whenever the Internet penetration is low? To check to what extent Internet may be a feasible option for on-line data-collection I ask myself three questions regarding the Internet penetration: 1. What is the Internet penetration figure in that Country, and how is it defined? 2. What is the total number of absolute users of the Internet in that country? 3. Out of those, which percentage connects to the Internet using a broadband connection? Let me further address these questions below by taking Slovenia and the Ukraine as an example.

1. Internet penetration figures The Internet penetration in Slovenia is 58% according to Eurostat. No figures are given for the Ukraine, so I use the Internet World Stats page with figures all countries in the world. On the Ukraine they report 12%. Since both Internet usage numbers come from various sources so we need to understand the exact definition before comparing: "ever used the Internet" versus "used in the past two weeks" makes quite a difference. Same counts for Internet penetration on a household level versus an individual level. I normally use the first rule of thumb to check feasibility of the method: whenever the general Internet usage is over 50% of total population in a country, I consider on-line a possible alternative for research on consumer goods and services. So with 58% in Slovenia I'd conclude we should be able to use on-line as a platform, but the Ukraine seems to be a challenge… But there's more to ask: 2. Absolute amount of Internet users Slovenia has a total population of 2.0 million people; the Ukraine has 45.8 million inhabitants. This should put the percentages into perspective: 58% of 2 million gives over 1 million of Internet users while 12% of 45 million makes the total available universe in the Ukraine 5.3 million! If the Internet penetration is lower than 50% in a country, but the total amount of Internet users is around 10 million or more, Internet may still be a valid as an alternative approach

The total universe of Internet users may still give us a viable sample population, but there is even more to consider: 3. Number of broadband connections This figure is important because broadband users tend to interact with the Internet differently than dial-up users. Dial up is all about transactional use of the web (mail, limited file transfer etc), whereas broadband users surf the Web for content, information and entertainment and they are more likely to engage in online commerce and tend to consequently be more Internet savvy. We must remember that we don’t undertake on-line research using the Internet at large, we using panels, and these panels are representative neither of the Internet population as a whole nor the general public.On-line panel members are predominantly broadband internet users because online research requires both frequent access, higher bandwidth and longer periods on-line which are less viable for dial up customers. As a rule of thumb, for online to be relevant for conducting research in your market, there needs to be (overall or geographically specific) a broadband penetration of at least 30%. It's actually 27% in Slovenia and for the Ukraine the figures are not given. So while no strict rules can be given, you now realize that you probably should not be focussing at the internet penetration percentage alone, the absolute amount of users and the number of broadband connections is also entering into the feasibility equation!

After analysing the Internet usage in a country, we will now have to check the availability of a good quality online respondent panel of sufficient size and with a high enough response rate, but these open a whole new set of questions to ask, perhaps more about these questions in a future post!

Drop me a note, or better: leave a comment if you want to ask me more!

1 comment:

  1. Emiel,

    While I cannot comment on Eastern Europe and Estonia or Ukraine in particular, I have had the opportunity to do some experiments in Latin America where Internet penetration rates are equivalent and sometimes even lower. What I learned there might be applicable here, to a certain extent.

    What is striking about Latin America is how the digital divide matches the class divide. While the middle and upper classes (A, B, C+) have access to the Internet in a way that is comparable to most Western countries, there is a huge gap with a lower class with limited access to most of the modern world's material goods, education, health and, of course, the Internet.

    While we have been very successful at running online surveys with some target groups that are more likely to be online (subscribers of satellite TV, B2B, movie goers to a certain extent), I would not recommend online sample for any survey aimed at a broader population. For the same reason, social and political polling are still questionable online even in the US.

    So, instead of applying a general 30% rule of thumb, for an entire country, my preference is to use a 50-60% rule of thumb applied to a specific target group.

    Of course, once you have looked at the Internet penetration, the next step is to look at available sample sources, because in the end what matters is your actual sampling frame.