Turning Dirt Roads into Information Highways
As we analyze the use of the concept of internet diffusion found in economics, political science, and public policy research in this paper, we find that the research on internet diffusion is often characterized by extensive errors in operationalization and measurement. These errors are not caused only by imperfections and inherent biases in commonly used datasets. Rather, they stem primarily from a conceptual stretching whereby the indicators used are rarely connected to the properties of the concept. The following facts account for the outcome:
- First, the nature of the internet and its diffusion is better understood as a network good, but researchers rely on data about internet connectivity as an independent good.
- Second, whereas the large international datasets report on internet hosts, internet users, and subscribers, these categories mean very different things from country to country and result in peculiar patterns in national ranking that are not generalizable.
- Third, there is no correlation between the different variables that have been used by scholars in measuring internet diffusion.
These factors result in inherent biases in measuring the diffusion outcomes; in addition, the use of different indicators considerably affects the primacy of a particular causal explanation over the alternatives. Researchers should develop explicit methodologies for analyzing internet diffusion. This avoids substituting a particular research technique for methodology, as well as using unrepresentative nominal variables