A study of Internet Penetration Percents of Africa using digital divide models

While per capital GDP and other incomewealth related variables play significant roles in explaining variations in the Internet Penetration Percents (IPPs), they often fail to explain some significant fractions of the variations. Studies then added other TelecommunicationICT and socio-cultural variables and see whether some subset of them can significantly increase the fractions of variations explained. In this study we pursue a different approach. By examining the XY graphs between the IPPs and GDP-PPP, we try to spot clusters of countries either above or below the regression lines. Next we create zero-one variables for each identified cluster and see if it is statistically significant. If that were the case, we test if one or more of ICT and socio-cultural variables can explain. If that fails, we conclude that a variable representing the cluster must be added to the model. Our approach will explain at least as much or more than the traditional approach.