Page 130 - Culture Technology Communication

National Level Culture and Global Diffusion 113

plaining adoption timing than were typical economic or infrastruc-
ture variables. Bivariate relationships of the predictors and START
found Newspapers, GDP, Teledensity, Gender Empowerment, and
International Call Cost to have the highest correlations. These pre-
dictors represent all three economic, infrastructure, and cultural
categories. When put into models based on empirical relationships
the most powerful model included Teledensity, International Call
Cost, and English Language Ability as predictors of adoption timing.
The strong predictive nature of Teledensity was expected, however
the role of International Call Cost is less intuitive. International
Call Cost was a measure of network centrality where countries with
higher costs for a three minute call to the US were considered less
central to the global communication network. Equating network
centrality with the cost of a call to the US is appropriate for a study
of this innovation as the Internet originated in the US. Although the
predictive strength of this variable was not expected, its negative re-
19
lationship with the dependent variable was. It should also be noted
that the high correlations of Newspapers and GDP with the depend-
ent variable did not translate into strong predictive variables. In
models with other variables these variables were easily dominated
by International Call Cost. The weaker power of the GDP variable
was surprising given findings from previous research and intuition.
The prominent role of International Call Cost was also seen in
the reduced set of variables. In this analysis, the highest bivariate
correlations between the predictors and START, based on the reduced
sample, were between START and Teledensity, International Call
Cost, and School Enrollment. International Call Cost and School En-
2
rollment together formed the most powerful model with R .546. The
strong role of education in the diffusion process is also of interest as
it provides further support for the significant role of education in
many forms of development.
The next part of the analysis examined the relationship be-
tween predictor variables and the growth of the Internet within
countries. Although this process attempts to find the best model to
explain growth within countries, this is not an implicit assumption
that the diffusion processes in all countries are the same. It is rec-
ognized that the drivers of intra-country access growth will differ
among countries. The models merely represent the most common
driving forces across countries, having a better fit in some countries
than others. The sample used for this analysis, countries having the
Internet for at least five years, was naturally higher in those vari-
ables that explain adoption. The bivariate relationships between
the independent variables and the dependent variable GROWTH