Page 273 - Kaleidoscope Academic Conference Proceedings 2024

Innovation and Digital Transformation for a Sustainable World

capability approach, which emphasizes individual rather ● Model 1 included the most significant variable
than household or communal units of analysis [3]. Other from each category according to the simple
potential variables like electricity, mobile data regression.
subscriptions, or broadband subscriptions were excluded as ● Model 2 focused exclusively on education
they are not recorded at the individual level. variables.
● Model 3 assessed economic variables.
The independent variables, referred to in this paper as ● Model 4 was optimized using Akaike Information
“markers of choice,” were identified from the literature as Criterion (AIC) and Bayes Information Criterion
significant determinants of internet use [18] and categorized (BIC) stepwise regressions to find the best model
under the Equitable and Enriching Internet Access fit.
Framework [19]. This categorization resulted in five groups 6. RESULTS
of markers: equality, education, economics, health, and
governance, with two proxy variables representing each 6.1 Summary of Simple Regression Models
category (see Table 1). These proxies were sourced from
reputable global data banks, including UNESCO, the The results from the simple regression models are
United Nations Population Division, the Human summarized in Table 2. All variables in the simple
Development Index (HDI), the International regression models were found to be statistically significant
Telecommunication Union (ITU), the World Health (p < 0.05).
Organization (WHO), and the World Bank Group, selecting
only the most recent complete data, ranging from 2020 to Table 2 – Simple Regression Summary Statistics
2023. The final dataset comprised complete records for 104
countries. Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1
Table 1 – Independent Variables Co-
Variable efficient Standard P-Value R 2
Error
Category Proxy Variable (Est)
School enrollment, secondary, female (% Female 0.60192 0.04609 < 2e-16 0.6257
gross) Schooling ***
Equality
Urban population (% of total population) Urban 0.73632 0.06774 < 2e-16 0.5367
Population ***
School enrollment, secondary (% gross)
Education Secondary 0.63907 0.04623 < 2e-16 0.652
Mean years of schooling (years) Education ***
Fixed-broadband price basket (2017 PPP $) Mean Years 5.6095 0.3783 < 2e-16 0.6831
Schooling ***
Economics
Gross national income (GNI) per capita
(2017 PPP $) ICT Price -0.19773 0.09126 0.0326 0.044
Basket *
Life expectancy at birth
7.067 < 2e-16
Health GNI (PPP) 6.963 e-05 0.5024
Maternal mortality ratio (modeled estimate, e-04 ***
per 100,000 live births)
Life 2.2595 0.1828 < 2e-16 0.5998
Political Stability and Absence of Expectancy ***
Violence/Terrorism (percentile rank)
Governance Maternal 1.33e-14
Regulatory Quality (percentile rank) Mortality -0.08973 0.009971 *** 0.4426
Political 1.86e-12
5. METHODS Stability 0.51304 0.06398 *** 0.3866
Data were analyzed using RStudio [20], where simple Regulatory 0.5936 0.0556 < 2e-16 0.5277
regression models were first applied to establish Quality ***
relationships between each independent variable (markers
of choice) and the dependent variable (internet use). This Educational variables, particularly Mean Years Schooling
technique was chosen for its ability to individually assess (see Figure 1), Secondary Education, and Female
the impact of each marker on internet use, providing a clear, Schooling, were the most significant predictors of internet
isolated view of each relationship. Subsequently, multiple use, with high R-squared values indicating strong
regression models were applied to explore the complexity correlations.
and interdependence of these factors:

– 229 –