Content uploaded by Rushambwa Tawonga
Author content
All content in this area was uploaded by Rushambwa Tawonga on Nov 28, 2022
Content may be subject to copyright.
1
A gendered analysis of the digital divide in South Africa: Considerations of
gender, demographics and the digital divide using National Incomes
Dynamics Wave 5 data.
Bonginkosi T Mkhize
1
, Tawonga Rushambwa
2
November 2022
Abstract
This paper is a contribution to the discussion of the gendered nature of the digital
divide in South Africa and its implications for socioeconomic welfare under
pervasive/broader digital transformation processes. The study uses a descriptive
analysis and three composite index variables measuring digital access,
socioeconomic conditions, and social support index. It also uses a number of
demographic characteristics to assess differences in the distribution of the digital
divide between males and females. Generally, males have been demonstrated to
perform better regarding digital access when compared to females. While gains in
education have a positive effect on females’ digital access, education access is
influenced by socioeconomic conditions, while increasingly higher levels of social
support particularly government support is associated with a low index of digital
performance. There is a need to investigate why public support networks have not
been instrumental in improving digital access to low socioeconomic households and
individuals.
Introduction
Digital transformation understood as the use of new digital technologies such as
artificial intelligence, big data, 3D printing, social media platforms, mobile
platforms, analytics, sensors, and embedded devices to enable major improvements
in business and work is argued to bring about a transformation of societies (Warner
and Wäger, 2019). A study conceptualized digital transformation as an ongoing
process of strategic renewal that uses advances in digital technologies to build
capabilities that transforms an organization’s business model, collaborative
approach, and culture (Walton et al., 2013). Such change is argued to result from the
transformation in business models, customer experience, streamlining of operations,
1
Bonginkosi T Mkhize, Doctor of Business Administration, University of Liverpool, Systems and Social
Change Research
2
Tawonga Rushambwa, MA Development Studies, Digital Transformation, and socioeconomic
development.
2
and creation of new businesses brought about by digital technology integration in
the economy and society (Warner and Wäger, 2019). A defining characteristic of
pervasive digital technology is the incorporation of digital capabilities into everyday
physical products such as adding software to books, clothing, home appliances, cars,
etc., providing novel functions that dramatically improve access, distribution, and
use (Yoo, Henfridsson and Lyytinen, 2010). From a firm-level perspective, dynamic
capabilities stem from strategic change as an ongoing process of continuous
learning. At the individual level, this implies individuals experiencing digital
transformation require access to continuous learning and development and
acquisition of skills and competencies to remain relevant within transforming
institutions and organizations.
At the micro-foundations of dynamic capabilities shaping digital transformation are
individuals, their skills, competencies, differences, and experiences that shape their
ability to participate, collaborate and contribute to the processes of digital
transformation. In future scenarios advanced by advocates of the fourth industrial
revolution, established and emerging digital technologies such as cloud computing,
computer hardware, sensors, machine learning, artificial intelligence (AI), the
internet of things (IoT), big data and analytics, and robotic automation are argued to
converge creating new cyber-physical systems transforming the execution of work
and mediating social and work interactions and the participation of individuals
(Pollitzer, 2018). At present digital transformation in Africa broadly is in its early
stages, however, the nature of transformation as determined by emerging
technologies requires digital skills and access to advanced digital technologies,
which can enable participation as well as shaping digital technologies for contextual
problem-solving of the technology users. The ability of individuals, to garner the
skills and competencies to adapt to digital transformation and its effects is therefore
vital to understanding socioeconomic outcomes under digital transformation.
In this study, we use the national incomes dynamics study data wave 5 to model the
influence of gender and other demographic factors on digital technology access and
the nature of the digital divide in South Africa. We argue that the digital divide
continues to persist in South Africa with layers observed in other studies such as
physical access and skills imbalances, but we also examine the influence of the
physical divide. The study concludes with the structuration theory that the general
population lack the resources to shape the technology to advance their welfare, due
to limitations of the physical divide and challenges in development of the needed
digital skills.
3
Giddens (1984) Structuration theory and the digital divide
In this study, we use the structuration theory advanced by Giddens (1984), which is
a generalized theory of social organisation. This theory focuses on the relationship
between individuals and their immediate society and its structures, which delimits
the context in which the individuals exercise agency (Dixon et al., 2014).
Structuration theory advances the idea that the mutual dependence of structure and
agency through active historical processes shapes the outcomes of social
organisation. Over time as structuration occurs as agents use various resources and
rules of their systems and by so doing they either reproduce or redesign the structural
principles that organized their activities in the first place (Golsorkhi et al., 2010).
Human agents draw on social structures in their decisions and subsequent actions
while simultaneously their actions serve to produce and reproduce the social
structure. According to Giddens (1984) agency has reference to the capabilities of
the people who make decisions, that is the possibility of individual action, for
example actions of governments in investing in shaping the digital environment in a
society, the provision of institutions and services and changing patterns of use of
those institutions and facilities by the changing populations (Jones and Karsten,
2008).
Agency then is not only the skills and will of the individuals, but the capacity to
control resources and diverting these to the implementation of the decisions within
existing structural rules, the rules and resources being the structural properties of
social systems (Golsorkhi et al., 2010). Giddens (1984) argued that resources
available to individuals are of two forms, allocative and authoritative resources.
Allocative resources involve command over objects and other material phenomena
while authoritative resources concern command over people. Individuals have more
capacity for agency the more structural resources they hold, that is they are able to
control the structural parameters of their society, and the more plural the rules they
are able to negotiate. Resources are a source of transformative power while plurality
affords discretion, that is the alternatives available to decision makers. The structure
within which agency is exercised does not just have a sense of constraint but it is
enabling as it furnishes the resources and the rules that guide agency. Structuration
theory as such admits of structural continuity while allowing for deliberate
innovation and change through the decisions and actions of individuals, the agency
(Golsorkhi et al., 2010). However, it is also important to note that change is possible
through access to resources while lack of access to resources can subject a group or
4
individuals to structural violence, in which the structures of society become barriers
to social mobility and welfare.
Giddens argued further that structure is dual, implying that the structural properties
of social systems are both the medium and the outcome of the decisions and practices
they recursively organize with the structure both constraining and enabling. Through
repeated collective action, the individuals reproduce the social structures. Thus, the
structuration theory presents the premise for understanding the relationship of the
individual to social structures that balances the understanding of power of inertia of
social institutions and the possibility of individuals to enact change. Since human
agents reproduces the social structure through their decisions, Giddens argued that
social structures can be transformed to improve the material welfare of individuals
who operate within the constraints that such structure impose (Lamsal, 2012).
The influence of technology on social practice depends on the actions of the social
agents and how these agents engage with it. Applying this understanding to the
digital divide, the development and shape of the divide is determined by the actions
of the agents in their interaction with digital technologies. However, their
interactions are determined by their individual capabilities, and how their social
structures shape the extent to which they are able to develop such capabilities. While,
individuals may be aware of digital technologies and the need for digital capabilities,
intervening constraints such as low socioeconomic status may limit the extent to
which they are able to develop the capabilities they need to close the digital gap.
Technology drives societal change and plays a role of the social construction of
gender through creating new possibilities of how gender roles can be performed in
a new area (Dixon et al., 2014). The existence of the digital divide and factors that
may perpetuate or reinforce it, may stall the technological transformation of society
and its positive effects on society such as opening opportunities for women’s
participation in previously male dominated industries etc. Through lack of effective
agency and resources to shape the structural underpinnings of society, resource
deprived groups may experience marginalisation and welfare loss as the economy
restructures through processes of digital transformation.
Review of literature
Various studies have discussed the existence of the digital divide in South Africa,
conceptualized as inequalities in access to technology due to various factors. A study
defined the digital divide as significant disparities between groups who have access
to digital technologies and those who do not (Hendricks and Olawale, 2022).
5
Another study argues conceptualized the divide as the digital skills gap and a
physical access to information technology gap, with these divides influencing and
reinforcing each other (Kularski and Moller, 2012). While the digital divide is
evident among groups, the gendered digital divide is argued to be pervasive referring
to unequal access to and inability to use ICTs between men and women (Saha and
Zaman, 2017).
In discussing the digital divide, two layers have been identified to be associated with
it, the first layer is concerned with physical access to ICT infrastructure, while the
second layer is concerned with unequal skills and knowledge of handling ICT
devices (Acharya, 2017). The third layer argued to be consequent on inequality of
users’ capacity to exploit ICT means influences expected outcomes is coined digital
outcome divide. Developing countries have been argued to be focused on the first
layer, installation of ICT infrastructure and affordable ICT services (Acharya, 2017).
In the 1990s, the digital divide was focused on access to infrastructure mean essential
to access, and thus computers were the key variable differentiating people’s access
or non-access. In the 21st century the concept was broadened revealing multiple
manifestations of the divide including access to digital devices, quality of internet
connection, knowledge and skills of users and nature of technology use (Riggins and
Dewan, 2005). The studies in this era focused on qualitative growth as measured by
access, digital skills and usage gaps. These studies argued that having access to
digital devices and the internet is not adequate to address the digital divide as the
digital divide is a compendium of interconnected social, economic and technological
considerations that influence internet access and utilization (Warschauer, 2003;
Paré, 2016). Another group of studies focused on the global disparities in internet
access between developed and developing countries. These studies argued for the
need to mitigate the horizontal disparity in internet access on a global scale. While
the digital divide in terms of access to skills and usage is appropriate in the context
of a limited number of developing countries but not for the majority of developing
countries which are lagging behind in ICT infrastructure. It was also observed that
while the digital access divide continued to be an issue on a global scale as the digital
divide was demonstrably narrowing in developed countries, howbeit widening in
many developing countries (Riggins and Dewan, 2005; Huang and Chen, 2010;
Huang, 2021).
While both group of studies are relevant to the present discussion, it is the second
group of studies whose perspective is too relevant for the issues we believe are key
in thinking about the welfare implications of digital transformation in the developing
6
countries. Developed countries have progressively shifted their focus on skills and
internet usage, the divide in internet access between developing and developed
countries is deepening due to the physical divide. In the second group of literature
this was observed that the gap in access between developing and developed countries
will be deep and wide in the long run due to the socioeconomic factors (Riggins and
Dewan, 2005). In recent studies, the first layer is still observed as characteristic as
technology increasingly shape political, economic and social spheres of society,
making the disparities between those with access and reap the positive externalities
associated with access and those who do not (Hendricks and Olawale, 2022).
In South Africa, the gender based digital divide is argued to be a reflection of social
inequality, with women being underrepresented in ICTs (Nesaratnam, Mamba and
Singh, 2018). Broadly studies have found that this digital gender divide is associated
with factors including barriers to access, affordability, lower levels of educational
attainment and technology illiteracy and sociocultural norms engendering digital
exclusion among females (OECD, 2018; James, 2022). At the global level, illiteracy
in the developing world has been shown to be a strong determinant of digital
exclusion in terms of access and use. It was observed in a study that 60% of adult
women are illiterate although differences do exist in countries. Another study argued
that welfare gains from digital access are connected with digital utilization in
productivity as there are various factors intervening between access and use,
principally, digital literacy, relevance and social relations between genders in the
household (Mumporeze and Prieler, 2017). A South African study focused on
behavioural/psychological barriers to digital access and use, and argued that
technologies engender negative emotions, fear and anxiety among due to existence
of norms of individual behaviour patterns. The study found that this psychological
effect was associated with age, employment and educational level and sustained the
digital divide through limiting access and skills (S. T. Faloye, S. Ranjeeth, and M.
A. Sonny Ako-Nai, 2022). Similar findings of psychological barriers were found in
the Rwandan study, alluding to the existence of social, economic and cultural factors
(Mumporeze and Prieler, 2017).
In a systematic literature review study, the digital divide was observed to be of a
different nature between developed and developing countries, with women in the
developing countries experiencing the adverse outcomes of the digital divide. While
the review argued that the primary causes were sociocultural factors, it also argued
for the need for further research to better understand how to address the sociocultural
factors affecting the gender digital divide, to understand why equal access does not
7
seem to result in equal use (Acilar and Sæbø, 2021). Several studies in the second
decade of the 21st century argued for the need for addressing long-existing
socioeconomic and socio-cultural inequalities which underpin social structuring, as
bridging the technological gaps may not necessarily address the real issue of the
digital divide (Nguyen, 2012; Harambam, Aupers and Houtman, 2013; Alcalá,
2014).
Methods
This study uses a descriptive quantitative research design in which we sought to
describe the gendered profile of the digital and the physical divide (socioeconomic
and demographic profile) of individuals and households. The choice of descriptive
quantitative design stemmed from the objective of this study to describe status of the
digital and physical divide across individuals in South Africa. We also intended to
trace the nature of the distribution of the digital and physical divide and the
implications for socioeconomic welfare under digital transformation.
Using the NIDS dataset (described below), three composite indexes were developed,
measuring the digital divide, socioeconomic status and social support and a
comparative analysis was undertaken on individuals according to their demographic
characteristics using one way analysis of variance (ANOVA), with post hoc analysis.
Through the descriptive design, the study was able to describe the nature of the
digital divide and its many faces and then assess its distribution across the population
and the implications of such distribution when the influence of demographic
variables, socioeconomic index and social index were included.
Data
The study uses the National Income Dynamics Study (NIDS) wave 5 (V1.0.0)
dataset published in 2018 by the Southern Africa Labour and Development Research
Unit at the University of Cape Town. The NIDS is a longitudinal panel dataset based
on surveys collecting data on income and expenditure and socioeconomic aspects of
individuals and households in South Africa over time. At the broadest level, the
NIDS data tracks variables measuring wealth creation through income and
expenditure and asset endowments, demographic dynamics, education and
employment dynamics, social capital, social services and intergenerational
transitions and development.
Variables
8
The key variables of analysis focused on how the physical conditions of the
population under consideration are associated with the layers of the digital divide
with particular focus on the dynamics between males and females in their various
demographic profiles. The key variables were socioeconomic conditions (measured
by a composite index composed of 17 attributes focusing on the material conditions
of the households), social influence (cultural and religious influence), access to
developmental opportunities (three attributes, dividing individuals into
communication active, working class, and Career oriented) and access to
infrastructure and digital index (measuring access to computers, smart phones,
internet access, satellite access—access to information and telephones). These
variables were created using tetrachoric correlation analysis based principal
components analysis. Principal components analysis was used to reduce dimension
on highly associated variables so that analysis can be based on unique variation
among the variables employed in the analysis.
Statistical Analysis
Stata software version 17.0 (Stata Corp, College Station Texas) was used for all the
statistical analyses reported in this study. The criterion for assessing statistical
significance was 5 percent for all bivariate and group analyses conducted. For
descriptive analysis, the mean, standard deviation, percentages and 95% confidence
intervals were calculated to assess the distribution of the variables across
demographic characteristics. Using tetrachoric correlations on binary variables, we
generated composite variables using principal components analysis on the
tetrachoric correlation matrix. To ensure that dimensions among variables were
reduced to the components that were highly correlated, we first carried out first stage
tetrachoric correlation analysis followed by rotated principal components setting
blanks to 30%. We then extracted the variables that reached the 30% minimum
correlation threshold and carried out second stage tetrachoric correlation analysis
and then predicted new variables based on the scores of the variables grouped under
the components. These new variables measuring digital access, socioeconomic
conditions, access to personal development opportunities, social influence and
language and computer literacy were used in the analysis together with demographic
variables, principally, age, gender and spatial location. The results of the findings
have been reported using graphical bar charts, and statistical tables showing
summary statistics. The use of tetrachoric correlations in carrying out principal
components analysis on binary transformed categorical variable, enable avoidance
of information loss due to truncation and also enabled assessment of true
9
associations among different variables. The values of sample size for individuals (N
= +28000) and households (N = +10000) was preserved in the estimation sample,
which shows that possible information loss or truncation of estimation sample size
was avoided. We therefore propose use of tetrachoric correlations for categorical
data analysis as competitive.
Findings
Digital access
In the analysis we assessed digital access measures of access to basic ICT
infrastructure such as telephones and computers, we also included smartphones
which have become relevant due to the proliferation of mobile based applications.
We also examined access to information such as news (satellite access) and levels
of computer literacy. The results of the analysis are presented in Table 1.
Table 1: Comparing the mean Digital Index over Education and Gender.
Education Levels
Gender
Key
Male
Female
No Schooling
Mean
Sd
N
2.59
(3.74)
687
2.72
(4.05)
860
Primary Education
Mean
Sd
N
2.89
(4.45)
1305
2.80
(4.71)
1358
Secondary Education
Mean
Sd
N
3.60
(5.28)
1644
3.02
(4.64)
1725
Certificate without Matric
Mean
Sd
N
3.69
(5.13)
274
3.85
(6.55)
186
Matric
Mean
Sd
N
4.08
(5.70)
941
3.88
(5.46)
933
Diploma with Matric
Mean
Sd
N
4.66
(6.18)
655
4.89
(6.36)
798
10
Higher Education
(University Degree)
Mean
Sd
N
5.65
(6.71)
361
5.89
(7.02)
395
The digital index was computed over (N = 12224) complete cases with a minimum
value of 0.00032 and a maximum value of 42.289. The distribution of the digital
index shown in figure 1 below shows that the majority of the sample population are
concentrated within the low range of the index, while the standard deviations (Sd)
in Table 1 above shows large dispersion of observations from the sample mean. The
large majority of South Africans generally have a very low mean digital index score,
while few have very high digital index scores, showing the disproportionate
distribution of access to digital assets, information and digital facilities.
Females with no schooling have a higher average digital index value than males with
no schooling. The number of females with no schooling is also larger than the
number of males. Looking at table 1, it can be observed that post matric education
is associated with higher mean values of the digital index for females than males,
which may be indicative of greater utilization of digital devices, access to
information and digital facilities for more educated women than males. Thus,
increasing access to education and training can be instrumental in improving the
profile of digital access for women and its productivity benefits. The analysis in
Table 1 shows that the digital index for women rises somewhat proportionately to
the changes in their associated levels of education. Furthermore, when we
decomposed the digital index, to reflect digital poverty, males had a small negative
index, while broadly females had a small positive mean index. The comparison in
Table 1A, showed that marginally males in South Africa generally perform better in
terms of digital access when compared to females.
Table 1A: Assessing Marginal Digital Poverty (Males versus Females)
Gender
Index of Estimated Digital Poverty
Mean
SD
Frequency
Male
-0.0001735
0.0070155
21529
Females
0.0011118
0.0754657
25452
11
Figure 1: Distribution of the Digital Index
Analysis of Variance
We ran analysis of variance between education levels and the digital index, gender
and the digital index and gender with each of the three components of the digital
index (digital assets, smartphone access and no digital access) and an interaction of
12
gender and education and digital index to test for statistical significance in the mean
distribution of the digital index across groups. The results presented in Appendix 1
revealed statistically significant differences between education levels and the digital
index, no statistically significant difference between gender and the overall digital
index however statistically significant differences for gender were found on the
components of the digital index. Finally, we found statistically significant
differences of the mean differences in the digital index for different categories of the
interaction between education and gender. The low mean for digital assets
component of the digital index, means that women have disproportionately low
access to digital devices such as computers, have low access to information and low
access to internet facilities when compared to males. While the differences were
significant, gender differences accounted for a significantly smaller share, less than
1% (0.0578%) of variation in the digital asset component. This shows that there are
other factors that influence access to digital assets although linked to gender
characteristics. Pairwise comparison in Appendix 1 shows that access to digital
assets, information and internet facilities is higher for females with higher levels of
education, particularly from matric to higher education.
Broadly in Table 2 below, the mean of the digital index rises in tandem with
increases in the level of education, and these differences are seen to be significant
between groups (males and females) and within groups (between females). These
findings suggest that ameliorating the digital divide will require improving access to
higher education and training for females, or reducing continuing barriers to higher
education for females in South Africa. The tables of statistics in Table 2 shows that
at levels of education from Matric to higher education, the mean of the digital index
for females becomes larger in absolute size than for males, with the number of
females in these higher education levels being higher than males.
Table 2: Digital Index over Education by Gender
Tabulation for Males
w5_best_edu
(Best Education)
Summary of Digital Index
Mean
Std. dev.
Freq.
No Schooling
2.5868586
3.7416704
687
Primary Education
2.8855783
4.4526392
1,305
Secondary Education
3.5967899
5.2825327
1,644
Certificate No Matric
3.6918101
5.1307682
274
Matric
4.0780915
5.7038239
941
Diploma with Matric
4.6626857
6.1797402
655
Higher Education
5.6469269
6.7085993
361
13
Total
3.6471133
5.2887814
5,867
Tabulation for Females
w5_best_edu
(Best Education)
Summary of Digital Index
Mean
Std. dev.
Freq.
No Schooling
2.7234369
4.0535063
860
Primary Education
2.8007488
4.713427
1,358
Secondary Education
3.0170293
4.6387911
1,725
Certificate No Matric
3.852317
6.5497177
186
Matric
3.8829733
5.4619456
933
Diploma with Matric
4.8876881
6.3604964
798
Higher Education
5.8922883
7.015923
395
Total
3.5039364
5.278908
6,255
Demographic Factors and Digital Index
Table 3: Digital Index over Age Groupings
Age
Groupings
Summary of Digital Index for
Males
Mean
Std. dev.
Freq.
Below 20 years,
2.8831775
4.556113
1,828
Above 21, below 30
3.7308303
5.4986226
1,364
Above 31, below 40
3.977997
5.3704006
970
Above 41, below 50
4.5697747
5.755256
605
Above 51, below 60
4.2058285
5.9842869
546
Above 61, not missing
4.0487537
5.4775633
511
Total
3.6655163
5.3046699
5,824
Age
Groupings
Summary of Digital Index for
Females
Mean
Std. dev.
Freq.
Below 20 years,
2.5820071
4.3105129
1,835
Above 21, below 30
3.7552104
5.1337625
1,231
Above 31, below 40
4.2580683
5.8790002
921
Above 41, below 50
4.5179326
5.8909552
759
Above 51, below 60
3.5106717
5.2892748
692
Above 61, not missing
3.4604298
5.864453
794
Total
3.5122601
5.2865843
6,232
14
In Table 3 the distribution of the average values of the digital index are assessed over
age categories. The digital index increases with an increase in the age of the
population for both males and females. Young males (up to 20 years) are more
digitally connected than their female counterparts (2.88 versus 2.58). Females
between the ages of 21 and 40 are more digitally connected when compared to their
male counterparts.
Figure 2: The physical divide (Socioeconomic conditions, social connectedness and
subjective wellbeing)
In figure 2 an assessment was made of the mean differences in the scores for
variables measuring the physical living conditions of males and females across their
respective population groups. The social index measured access to public support
networks, public services group membership and general preference towards
contextual environment at time of survey. For both males and females, the social
index was relatively at the same level with government support being the most
influential component of social support, influencing 11% of the overall variation in
the index. Figure 2 also shows that males have better scores on subjective wellbeing
than females. The socioeconomic conditions index examined the wealth conditions
15
of the individuals within their households. The household structure and its design
were used as measures of socioeconomic performance. Generally, Africans showed
poor socioeconomic performance when compared with other population groups
although the performance between African males and females does not differ by a
larger margin. We used a dichotomized variable of digital performance with 3 levels
low digital index, moderate digital index and high digital index and employed
ANOVA to assess whether socioeconomic and social connectedness differed for
each level for the levels of the dichotomized variables.
Socioeconomic performance and its distribution
Figure 3: Socioeconomic Performance index
The left skewed distribution of the socioeconomic performance index, shows that on
average individuals in South Africa experience limiting socioeconomic conditions.
When we decompose socioeconomic performance across gender and population
groups (Table 4a), African females have the lowest mean value for socioeconomic
performance, implying that they comparatively more sharply experience the
reversals of limiting socioeconomic conditions. Stable housing conditions and better
material welfare accords individuals with time, to invest in acquiring digital skills,
or advancing their education and training prospects. As table 4b shows, better
16
socioeconomic performance is associated with high levels of digital access and
digital technology utilization.
Table 4a: Decomposition of Socioeconomic performance over Population group and
gender.
Population
Group
Male Female
Total
African
23.008228 22.405571
µ
22.696266
16.455336 13.502007
Sd
15.001444
4005 4298
N
8303
Coloured
27.886274 28.367201
µ
28.141361
12.944253 14.384054
Sd
13.723635
610 689
N
1299
Indian/Asian
33.242765 31.110823
µ
32.171438
14.251052 12.43226
Sd
13.376961
99 100
N
199
White
32.173061 33.192116
µ
32.721877
11.739509 14.206988
Sd
13.128295
383 447
N
830
Total
24.479477 24.176383
µ
24.321701
16.001413 14.08548
Sd
15.034615
5097 5534
N
10631
Table 4b: Socioeconomic conditions and digital access
Digital Index
Dichotomized
Index Categories
Summary of Socioeconomic Conditions
Mean
Std. dev. Freq.
High Digital Index
28.076053
14.950161 4,208
Moderate Digital
Index
21.808121
15.009098 5,181
Low Digital Index
22.300591
13.277579 1,259
Total
24.343384
15.095488 10,648
Digital Access
Index Categories
Summary of Household monthly - full
imputations income
Mean Std. dev.
Freq.
High Digital Index
19108.383 48855.114
4,208
17
Moderately Low
5255.3522 5950.4916
5,181
Low Digital Index
6568.8357 11165.584
1,259
Total
10885.257 31928.451
10,648
In Table 4, we observe that better socioeconomic conditions are associated with
better digital access. In computation of the socioeconomic performance index,
components used measured long-term wealth and focused on housing
infrastructures. This shows that the physical conditions of people, such as access to
stable housing, safe living conditions, sanitation and clean energy have a positive
influence on the digital performance score. The implicit cost of digital asset access
is income, this is demonstrated in Table 4 (b) above. Household with large average
household incomes, demonstrate higher performance score in digital access.
Table 5: Social Service Access and Digital Performance
Digital
Index Categories
Summary of Social Access Index
Mean
Std. dev.
Freq.
High Digital Index
26.582403
47.432446
4,208
Moderately Low Index
44.757713
60.639317
5,181
Low Digital Index
35.948359
43.831853
2,657
Total
36.465486
53.428218
12,046
The social access index measured available support services to individuals and
households such as social services, group membership, access to social infrastructure
and institutions. There is an inverse relationship between digital access index and
social access index, higher mean values of the social access index are associated
with lower values of the digital access index. This demonstrates that the higher the
dependence of individuals on external support, the lower they are likely to invest in
digital access, maybe due to the directed nature of the external support, or its size.
Given the higher component of government support in the social access index, it can
be concluded that either household receiving support are grossly impoverished or
technology does not mediate aspects of their lives to a significant extent. This might
require further investigation.
Discussion
18
Our study identified the existence of various aspects of the digital divide, as physical
access identified by access to digital assets and skills, although we limited our
analysis to physical access. We found the level of social performance as important
in determining digital access. We also observed that generally digital access
improved with increase in levels of education. The review of literature showed that
gender and other demographic factors have an influence on digital technology access
and the nature of the digital divide in South Africa (Kularski and Moller, 2012; Saha
and Zaman, 2017; Hendricks and Olawale, 2022). While education is an important
in improving digital performance among households and individuals, this is
indicative of the phenomenon that digital access among individuals and households
in South Africa is linked to its productivity effect. However, access to education is
linked to the social conditions of individuals and households, with basically
individuals from the African population group experiencing very poor
socioeconomic performance on average. Government support through grants has
been shown to not influence digital access, given the high levels of the social index
being associated with low mean values of digital performance. A study of digital
inequalities in the Netherlands, also found socioeconomic determinants as crucial
even when internet infrastructure is provided (van Deursen and van Dijk, 2019).
The results of this study indicated that the digital divide continues to persist in South
Africa with layers as observed in other studies such as physical access and skills
imbalances. This finding is consistent with that of Squicciarini (2018) who refers to
the divide as an emerging innovative form of inequality in South Africa that is based
on digital skills, access, and gender. Furthermore, these results reflect those of
Mariscal et al. (2020) who found that despite the remarkable progress made in digital
access, significant inclusion of women in the digital revolution was not yet evident.
On the question of status of the digital and physical divide across individuals in
South Africa, this study concurred with the structuration theory that the general
population lack the resources to shape the technology to advance their welfare, due
to limitations of the physical divide and challenges in development of the needed
digital skills (Takyar, 2022; Benjamin, 2021). In South Africa, because of the
dominant patriarchal system in society, the digital transformation has enhanced the
gender-related inequality significantly (Saha & Zaman, 2017; Antonio & Tuffley,
2014). Consequently, women’s marginalization in digital technologies continues to
be evident even in South Africa’ education system (Hendricks & Olawale, 2022).
Moreover, according to Kiss & Abdullatif (2019), in South Africa there are notable
19
differences between those who have digital access, skills, and capacity to benefit
them and those without.
The aggravated nature of the entrenched physical divide in South Africa can be
observed in that those more dependent on social support, have veery low levels of
digital access. This shows that in South Africa, social support particularly
government grants are directed towards the more impoverished segment of the
population, who are also at an aggravated risk of marginalisation in the case of
pervasive digital transformation. However, this result has not previously been
described in previous studies. This intriguing finding could be attributed to the
directed nature of the external Government support, or its size.
Other studies also focus on the entrenched nature of the dgital divide beyond digital
access (S. T. Faloye, S. Ranjeeth, and M. A. Sonny Ako-Nai, 2022), however this
and other studies have not focused on the gendered nature of the divide and its
socioeconomic foundations in South Africa. Therefore, this study provides greater
insight in the status and nature of digital divide in South Africa. Furthermore, the
data is generally representative of the phenomenon across South Africa at the
national level. Further research is required to establish why individuals who are
highly dependent on individuals Government support have a lower likelihood to
invest in digital access.
Conclusion
This study set out to describe the status of the digital and physical divide across
individuals in South Africa. The study also intended to trace the nature of the
distribution of the digital and physical divide. Furthermore, the study aimed to
determine the implications for socioeconomic welfare under digital transformation.
As detailed in the discussion section the findings clearly indicate that the digital
divide continues to persist in South Africa with layers as observed in other studies
such as physical access and skills imbalances. The second major finding was that the
general population lack the resources to shape the technology to advance their
welfare, due to limitations of the physical divide and challenges in development of
the needed digital skills. Nevertheless, the most significant finding from this study
was that in South Africa, because of the dominant patriarchal system in society, the
digital transformation has enhanced the gender-related inequality significantly with
consequent women’s marginalization in digital technologies even in the education
system. Furthermore, another intriguing and unanticipated finding to emerge from
this study was that the higher the dependence of individuals on external support,
20
such as Government support, the lower they are likely to invest in digital access.
Taken together, these results suggest that there is a link between South Africa’s
deeply entrenched society–wide patriarchal system and one of the widest gender-
related inequality in digital transformation. The insights gained from this study may
be of assistance to a growing body of evidence that may help us to design new
informed strategies to deal with the digital transformation divide gap.
One of the strengths of this study is that it represents a comprehensive examination
of the whole of South Africa thus providing a more representative national picture
of the status and nature of digital divide. Therefore, this study lays the groundwork
for future research to establish why individuals who are highly dependent on
individuals Government support have a lower likelihood to invest in digital access.
It is unfortunate that this study did not include transgender individuals in its
gendered analysis of the digital divide in South Africa. Notwithstanding this
limitation this work offers valuable insights into the impact of gender digital
inequality in South Africa. A greater focus on transgender individuals could produce
broad all-encompassing findings that account more for the analysis of gender digital
inequality in South Africa. It is recommended that this information be used by
government policy makers to develop targeted interventions aimed at reducing the
disparity in gender digital access and skills in South Africa.
References
Acharya, B.B. (2017) ‘Conceptual evolution of the digital divide: A systematic
review of the literature over a period of five years (2010-2015)’, World of Media.
Journal of Russian Media and Journalism Studies, (7), pp. 41–74.
Acilar, A. and Sæbø, Ø. (2021) ‘Towards understanding the gender digital divide: a
systematic literature review’, Global Knowledge, Memory and Communication,
ahead-of-print(ahead-of-print). Available at: https://doi.org/10.1108/GKMC-09-
2021-0147.
Alcalá, L.A. (2014) ‘Media literacy for older people facing the digital divide: The e-
inclusion programmes design’, Comunicar. Media Education Research Journal,
22(1).
van Deursen, A.J. and van Dijk, J.A. (2019) ‘The first-level digital divide shifts from
inequalities in physical access to inequalities in material access’, New Media &
21
Society, 21(2), pp. 354–375. Available at:
https://doi.org/10.1177/1461444818797082.
Dixon, L.J. et al. (2014) ‘Gendered Space: The Digital Divide between Male and
Female Users in Internet Public Access Sites*’, Journal of Computer-Mediated
Communication, 19(4), pp. 991–1009. Available at:
https://doi.org/10.1111/jcc4.12088.
Golsorkhi, D. et al. (2010) Cambridge Handbook of Strategy as Practice.
Cambridge University Press.
Harambam, J., Aupers, S. and Houtman, D. (2013) ‘The contentious gap: From
digital divide to cultural beliefs about online interactions’, Information,
Communication & Society, 16(7), pp. 1093–1114.
Hendricks, W. and Olawale, B.E. (2022) BRIDGING THE GENDER-BASED
DIGITAL DIVIDE: EMPOWERMENT OF WOMEN THROUGH ICT. Conference:
Tenth Pan-Commonwealth Forum on Open Learning (PCF10). Available at:
https://doi.org/10.56059/pcf10.9136.
Huang, C.-Y. and Chen, H.-N. (2010) ‘Global digital divide: A dynamic analysis
based on the bass model’, Journal of Public Policy & Marketing, 29(2), pp. 248–
264.
Huang, E.Y. (2021) ‘Ocean Media: Digital South China Sea and Gilles Deleuze’s
Desert Islands’, Verge: Studies in Global Asias, 7(2), pp. 177–203. Available at:
https://doi.org/10.5749/vergstudglobasia.7.2.0177.
James, J. (2022) Gender, Internet Use, and Covid-19 in the Global South: Multiple
Causalities and Policy Options. Springer Nature.
Jones, M.R. and Karsten, H. (2008) ‘Giddens’s Structuration Theory and
Information Systems Research’, MIS Quarterly, 32(1), pp. 127–157. Available at:
https://doi.org/10.2307/25148831.
Kularski, C. and Moller, S. (2012) ‘The digital divide as a continuation of traditional
systems of inequality’, Sociology, 5151(15), pp. 1–23.
Lamsal, M. (2012) ‘The structuration approach of Anthony Giddens’, Himalayan
Journal of Sociology and Anthropology, 5, pp. 111–122.
22
Mumporeze, N. and Prieler, M. (2017) ‘Gender digital divide in Rwanda’,
Telematics and Informatics, 34(7), pp. 1285–1293. Available at:
https://doi.org/10.1016/j.tele.2017.05.014.
Nesaratnam, S., Mamba, T.N. and Singh, J.S.K. (2018) ‘A Study on the Gender
Digital Divide: Overcoming the Under-Representation of South African Women in
the IT Sector’, Business Management and Strategy, 9(2), pp. 151–168. Available at:
https://ideas.repec.org/a/mth/bmsmti/v9y2018i2p151-168.html (Accessed: 7
November 2022).
Nguyen, A. (2012) ‘The digital divide versus the “digital delay”: Implications from
a forecasting model of online news adoption and use’, International Journal of
Media & Cultural Politics, 8(2–3), pp. 251–268.
OECD (2018) ‘Bridging the digital gender divide: include, upskill, innovate.’
Organisation for Economic Co-operation and Development. Available at:
http://www.oecd.org/going-digital/bridging-the-digital-gender-divide.pdf.
Paré, D. (2016) ‘The digital divide: why the ‘the’is misleading’, in Human rights in
the digital age. Routledge-Cavendish, pp. 99–112.
Pollitzer, E. (2018) ‘Creating a better future’, Journal of International Affairs, 72(1),
pp. 75–90.
Riggins, F.J. and Dewan, S. (2005) ‘The digital divide: Current and future research
directions’, Journal of the Association for information systems, 6(12), p. 4.
S. T. Faloye, S. Ranjeeth, and M. A. Sonny Ako-Nai (2022) ‘Impact of
Technophobia on the Digital Divide. A Preliminary Case Study in the Eastern Cape
Province of South Africa’, in 2022 IST-Africa Conference (IST-Africa). 2022 IST-
Africa Conference (IST-Africa), pp. 1–11. Available at:
https://doi.org/10.23919/IST-Africa56635.2022.9845655.
Saha, S.R. and Zaman, M.O. (2017) ‘Gender digital divide in higher education: a
study on University of Barisal, Bangladesh’, IOSR Journal of Humanities and Social
Science, 22(01), pp. 11–17.
Walton, P. et al. (2013) ‘A digital inclusion: Empowering all Australians’, Journal
of Telecommunications and the Digital Economy, 1(1), p. 9.1-9.17.
23
Warner, K.S. and Wäger, M. (2019) ‘Building dynamic capabilities for digital
transformation: An ongoing process of strategic renewal’, Long range planning,
52(3), pp. 326–349.
Warschauer, M. (2003) ‘Dissecting the" digital divide": A case study in Egypt’, The
information society, 19(4), pp. 297–304.
Yoo, Y., Henfridsson, O. and Lyytinen, K. (2010) ‘Research commentary—the new
organizing logic of digital innovation: an agenda for information systems research’,
Information systems research, 21(4), pp. 724–735.
Appendix 1: ANOVA POST HOC ANALYSIS.
Pairwise comparisons of means with equal
variances Over: EducationBins Gender
Number of comparisons
EducationBins#Gender 91
Tukey
Tukey
DigitalAssets
Contrast
Std. err.
t P>t
[95% conf.
interval]
EducationBins#Gender
(No Schooling#Female) vs (No
Schooling#Male)
-
.0062272
.0078524
-0.79
1.000
-.0325613
.0201068
(Primary Education#Male) vs (No
Schooling#Male)
.0049999
.007349
0.68
1.000
-.0196458
.0296456
(Primary Education#Female) vs (No
Schooling#Male)
.0004491
.0072273
0.06
1.000
-.0237886
.0246867
(Secondary Education#Male) vs (No
Schooling#Male)
.0598848
.007496
7.99
0.000
.0347461
.0850236
(Secondary Education#Female) vs (No
Schooling#Male)
.036512
.0071996
5.07
0.000
.0123674
.0606566
(Certificate No Matric#Male) vs (No
Schooling#Male)
.1504639
.0145669
10.33
0.000
.1016121
.1993156
(Certificate No Matric#Female) vs (No
Schooling#Male)
.0845979
.0151541
5.58
0.000
.0337767
.135419
24
(Matric#Male) vs (No Schooling#Male)
.1860129
.0093232
19.95
0.000
.1547464
.2172795
(Matric#Female) vs (No Schooling#Male)
.1401625
.0088256
15.88
0.000
.1105648
.1697602
(Diploma with Matric#Male) vs (No
Schooling#Male)
.292505
.0111399
26.26
0.000
.255146
.329864
(Diploma with Matric#Female) vs (No
Schooling#Male)
.2061542
.0097783
21.08
0.000
.1733615
.238947
(Higher Education#Male) vs (No
Schooling#Male)
.4664487
.015121
30.85
0.000
.4157387
.5171586
(Higher Education#Female) vs (No
Schooling#Male)
.4028489
.0142196
28.33
0.000
.355162
.4505359
(Primary Education#Male) vs (No
Schooling#Female)
.0112272
.0068173
1.65
0.934
-.0116355
.0340899
(Primary Education#Female) vs (No
Schooling#Female)
.0066763
.006686
1.00
0.999
-.015746
.0290986
(Secondary Education#Male) vs (No
Schooling#Female)
.0661121
.0069756
9.48
0.000
.0427188
.0895054
(Secondary Education#Female) vs (No
Schooling#Female)
.0427392
.006656
6.42
0.000
.0204176
.0650608
(Certificate No Matric#Male) vs (No
Schooling#Female)
.1566911
.014306
10.95
0.000
.1087142
.204668
(Certificate No Matric#Female) vs (No
Schooling#Female)
.0908251
.0149036
6.09
0.000
.0408443
.1408059
(Matric#Male) vs (No Schooling#Female)
.1922402
.0089102
21.58
0.000
.1623589
.2221215
(Matric#Female) vs (No Schooling#Female)
.1463897
.0083881
17.45
0.000
.1182594
.1745201
(Diploma with Matric#Male) vs (No
Schooling#Female)
.2987323
.0107966
27.67
0.000
.2625247
.3349398
(Diploma with Matric#Female) vs (No
Schooling#Female)
.2123815
.0093853
22.63
0.000
.1809068
.2438562
(Higher Education#Male) vs (No
Schooling#Female)
.4726759
.0148698
31.79
0.000
.4228081
.5225437
(Higher Education#Female) vs (No
Schooling#Female)
.4090762
.0139522
29.32
0.000
.3622858
.4558666
(Primary Education#Female) vs (Primary
Education#Male)
-
.0045509
.0060868
-0.75
1.000
-.0249637
.015862
(Secondary Education#Male) vs (Primary
Education#Male)
.0548849
.0064035
8.57
0.000
.0334099
.0763599
(Secondary Education#Female) vs (Primary
Education#Male)
.031512
.0060538
5.21
0.000
.0112098
.0518143
(Certificate No Matric#Male) vs (Primary
Education#Male)
.1454639
.014036
10.36
0.000
.0983926
.1925353
(Certificate No Matric#Female) vs (Primary
Education#Male)
.0795979
.0146445
5.44
0.000
.0304857
.1287102
25
(Matric#Male) vs (Primary Education#Male)
.181013
.0084698
21.37
0.000
.1526085
.2094176
(Matric#Female) vs (Primary
Education#Male)
.1351625
.0079187
17.07
0.000
.1086061
.161719
(Diploma with Matric#Male) vs (Primary
Education#Male)
.2875051
.0104361
27.55
0.000
.2525063
.3225039
(Diploma with Matric#Female) vs (Primary
Education#Male)
.2011543
.0089683
22.43
0.000
.171078
.2312306
(Higher Education#Male) vs (Primary
Education#Male)
.4614487
.0146102
31.58
0.000
.4124516
.5104459
(Higher Education#Female) vs (Primary
Education#Male)
.397849
.0136752
29.09
0.000
.3519876
.4437104
(Secondary Education#Male) vs (Primary
Education#Female)
.0594358
.0062635
9.49
0.000
.0384303
.0804412
(Secondary Education#Female) vs (Primary
Education#Female)
.0360629
.0059055
6.11
0.000
.016258
.0558679
(Certificate No Matric#Male) vs (Primary
Education#Female)
.1500148
.0139727
10.74
0.000
.1031558
.1968738
(Certificate No Matric#Female)
vs
(Primary Education#Female)
.0841488
.0145839
5.77
0.000
.0352401
.1330576
(Matric#Male) vs (Primary
Education#Female)
.1855639
.0083645
22.18
0.000
.1575126
.2136151
(Matric#Female) vs (Primary
Education#Female)
.1397134
.007806
17.90
0.000
.1135352
.1658916
(Diploma with Matric#Male) vs (Primary
Education#Female)
.292056
.0103508
28.22
0.000
.2573433
.3267686
(Diploma with Matric#Female) vs (Primary
Education#Female)
.2057052
.0088689
23.19
0.000
.1759623
.235448
(Higher Education#Male) vs (Primary
Education#Female)
.4659996
.0145494
32.03
0.000
.4172064
.5147928
(Higher Education#Female) vs (Primary
Education#Female)
.4023999
.0136102
29.57
0.000
.3567565
.4480433
(Secondary Education#Female) vs
(Secondary Education#Male)
-
.0233729
.0062315
-3.75
0.013
-.0442708 -
.0024749
(Certificate No Matric#Male) vs (Secondary
Education#Male)
.090579
.0141135
6.42
0.000
.0432476
.1379104
(Certificate No Matric#Female)
vs
(Secondary Education#Male)
.024713
.0147189
1.68
0.924
-.0246484
.0740745
(Matric#Male) vs (Secondary
Education#Male)
.1261281
.0085977
14.67
0.000
.0972947
.1549615
(Matric#Female) vs (Secondary
Education#Male)
.0802776
.0080554
9.97
0.000
.0532631
.1072922
26
(Diploma with Matric#Male) vs (Secondary
Education#Male)
.2326202
.0105402
22.07
0.000
.1972725
.2679679
(Diploma with Matric#Female) vs
(Secondary Education#Male)
.1462694
.0090892
16.09
0.000
.1157878
.176751
(Higher Education#Male) vs (Secondary
Education#Male)
.4065638
.0146847
27.69
0.000
.3573168
.4558108
(Higher Education#Female) vs (Secondary
Education#Male)
.3429641
.0137548
24.93
0.000
.2968359
.3890923
(Certificate No Matric#Male)
vs
(Secondary Education#Female)
.1139519
.0139583
8.16
0.000
.0671409
.1607628
(Certificate No Matric#Female)
vs
(Secondary Education#Female)
.0480859
.0145701
3.30
0.059
-.0007768
.0969486
(Matric#Male) vs (Secondary
Education#Female)
.149501
.0083405
17.92
0.000
.1215301
.1774718
(Matric#Female) vs (Secondary
Education#Female)
.1036505
.0077803
13.32
0.000
.0775585
.1297425
(Diploma with Matric#Male) vs (Secondary
Education#Female)
.255993
.0103315
24.78
0.000
.2213453
.2906408
(Diploma with Matric#Female)
vs
(Secondary Education#Female)
.1696422
.0088463
19.18
0.000
.1399752
.1993093
(Higher Education#Male) vs (Secondary
Education#Female)
.4299367
.0145357
29.58
0.000
.3811897
.4786837
(Higher Education#Female) vs (Secondary
Education#Female)
.366337
.0135955
26.95
0.000
.3207429
.411931
(Certificate No Matric#Female)
vs
(Certificate No Matric#Male)
-.065866
.0193041
-3.41
0.042
-.1306046 -
.0011273
(Matric#Male) vs (Certificate No
Matric#Male)
.0355491
.0151633
2.34
0.522
-.0153027
.0864008
(Matric#Female) vs (Certificate No
Matric#Male)
-
.0103014
.0148625
-0.69
1.000
-.0601444
.0395416
(Diploma with Matric#Male) vs (Certificate
No Matric#Male)
.1420412
.0163432
8.69
0.000
.0872324
.1968499
(Diploma with Matric#Female)
vs
(Certificate No Matric#Male)
.0556904
.0154472
3.61
0.022
.0038862
.1074945
27
(Higher Education#Male) vs (Certificate No
Matric#Male)
.3159848
.0192781
16.39
0.000
.2513334
.3806362
(Higher Education#Female) vs (Certificate
No Matric#Male)
.2523851
.0185795
13.58
0.000
.1900766
.3146936
(Matric#Male) vs (Certificate No
Matric#Female)
.1014151
.0157283
6.45
0.000
.0486685
.1541616
(Matric#Female) vs (Certificate No
Matric#Female)
.0555646
.0154385
3.60
0.022
.0037899
.1073393
(Diploma with Matric#Male)
vs
(Certificate No Matric#Female)
.2079071
.0168687
12.33
0.000
.151336
.2644783
(Diploma with Matric#Female)
vs
(Certificate No Matric#Female)
.1215563
.0160022
7.60
0.000
.0678911
.1752216
(Higher Education#Male) vs (Certificate No
Matric#Female)
.3818508
.0197256
19.36
0.000
.3156987
.4480029
(Higher Education#Female)
vs
(Certificate No Matric#Female)
.3182511
.0190434
16.71
0.000
.2543868
.3821153
(Matric#Female) vs (Matric#Male)
-
.0458505
.0097786
-4.69
0.000
-.0786442 -
.0130567
(Diploma with Matric#Male) vs
(Matric#Male)
.1064921
.0119092
8.94
0.000
.0665533
.1464308
(Diploma with Matric#Female) vs
(Matric#Male)
.0201413
.0106464
1.89
0.831
-.0155627
.0558452
(Higher Education#Male) vs (Matric#Male)
.2804357
.0156963
17.87
0.000
.2277963
.3330752
(Higher Education#Female) vs (Matric#Male)
.216836
.0148299
14.62
0.000
.1671022
.2665698
(Diploma with Matric#Male) vs
(Matric#Female)
.1523425
.0115237
13.22
0.000
.1136963
.1909887
(Diploma with Matric#Female) vs
(Matric#Female)
.0659917
.0102134
6.46
0.000
.0317398
.1002437
(Higher Education#Male) vs (Matric#Female)
.3262862
.0154059
21.18
0.000
.2746206
.3779518
(Higher Education#Female) vs
(Matric#Female)
.2626865
.0145222
18.09
0.000
.2139845
.3113884
(Diploma with Matric#Female) vs (Diploma
with Matric#Male)
-
.0863508
.0122687
-7.04
0.000
-.1274953 -
.0452063
(Higher Education#Male) vs (Diploma with
Matric#Male)
.1739437
.0168389
10.33
0.000
.1174724
.2304149
28
(Higher Education#Female) vs (Diploma with
Matric#Male)
.1103439
.0160344
6.88
0.000
.0565708
.1641171
(Higher Education#Male) vs (Diploma with
Matric#Female)
.2602944
.0159708
16.30
0.000
.2067344
.3138545
(Higher Education#Female) vs (Diploma with
Matric#Female)
.1966947
.0151202
13.01
0.000
.1459875
.2474019
(Higher Education#Female) vs (Higher
Education#Male)
-
.0635997
.019017
-3.34
0.051
-.1273756
.0001761
Source: own calculations using NIDS Wave 5 Data
