Methodological challenges of digital divide measurements

The term digital divide has been used for almost a decade and typically relates to sociodemographic differences in the use of information and communication technology. However, the corresponding measurement is still relatively imprecise. Very often it is simply reduced to comparisons of Internet penetration rates. This article extends the measurements above the usual bivariate comparisons. Within this context, three essential approaches are presented and critically evaluated. First, loglinear modeling is used to address the interactions among the factors affecting the digital divide. Second, compound measures (i.e., the Digital Divide Index) that integrate a number of variables into a single indicator are discussed. Third, time-distance methodology is applied to analyze changes in the digital divide. The article argues that these approaches often yield entirely different conclusions compared to simple bivariate analysis. The examples are presented as a general warning against an oversimplified methodological approach to digital divide studies.