Page 124 - Culture Technology Communication

National Level Culture and Global Diffusion 107

Examining the relative power of variables within models by the
15
standardized betas the following results are obtained (table 4). In
models 1a, 1b, 2b, 5, and 6 it can be seen that International Call Cost
is a stronger predictor than Newpapers, English Language Ability
(TOEFL), GDP per Capita, PCs per thousand, and Links. Models 3
and 4 respectively show that Teledensity and Gender Empowerment
are stronger predictors than International Call Cost. A direct com-
parison of Teledensity and Gender Empowerment was not possible
due to multicollinearity (r .689). However, the partial correlations
of each variable with the dependent variable controlling for one an-
other would indicate Teledensity has a stronger relationship with
START than does Gender Empowerment. 16 Model 2a is of interest
because it contradicts the other models showing School Enrollment
to have greater explanatory power than International Call Cost.
Model 2b shows the English Language Ability factor to be almost as
powerful as that of GDP, with respective standardized betas of .185
and .201. Therefore, the top three overall predictors are Teledensity
2
2
( R .526), Gender Empowerment ( R .468), and International
2
Call Cost ( R .438).
METHOD 2
The results obtained from the second method are as follows (table 5).
In method 2 a sub-sample of fifty-five countries was used. The sub-
sample includes both developed and developing countries, those with
old and recent adoptions, and represent a variety of cultural back-
grounds. For the reduced set of countries, Teledensity, International
Call Cost, and School Enrollment are the top three most highly cor-
related variables with the dependent variable, START. These vari-
ables were combined with other variables based on their own
correlation with the dependent variable and their relationship to one
another. This process produced the following models and results.
With this reduced sample the model with the highest explana-
2
tory power (R .546) includes International Call Cost and School
Enrollment as the predictors. The strongest model in the first sec-
tion includes Teledensity, International Call Cost, and English
Language Ability (TOEFL). Unfortunately, in this smaller sample of
countries the Teledensity and International Call Cost variables are
too highly correlated (r .6) to test this as a model. However, as with
the previous analysis, when Teledensity was combined with
TOEFL, the only variable able to be paired with Teledensity, Tele-
density was the stronger predictor. Also, International Call Cost is