You are on page 1of 8

SPECIAL ARTICLE

From Mobile Access to Use


Evidence of Feature-level Digital Divides in India

Jang Bahadur Singh, M Vimalkumar

I
The digital divide is the disparity between individuals nformation and communications technologies (ICTs) facilitate
with respect to access to information and the economic development of the poor and marginalised
by improving their access to education, healthcare, and
communication technologies. The growing prevalence
financial services (Waverman et al 2005; O’Riain 2004;
of mobile phones in India is often linked to phones Steinmuller 2001). Thus, it is important to bridge the digital
becoming access points to various government divide—the inequality between the technological “haves” and
schemes and services. However, ICTs have various “have nots” (Dewan and Riggins 2005; OECD 2001; Wei et al
2011)—and promote equitable access to ICTs such as personal
features that are not uniformly operated by different
computers and mobile phones.
users. The use of mobile phones is examined Over the last decade, India has witnessed an exponential
using micro-level data to highlight how the growth in the penetration of mobile phones, and according to
socio-demographic characteristics of individuals recent statistics, there are 88 mobile phone connections for
every 100 individuals (TRAI 2017). Mobile phones are a major
(age, gender, literacy, etc) influence their engagement
channel for accessing the internet in India—approximately
with the various features of a mobile phone. 81% of internet users in India access it using smartphones
(Neeraj 2016). Although smartphone penetration is still low,
its growth has been robust due to the recent availability of
low-cost handsets (Euromonitor 2017). This has led to growing
optimism that the digital divide can be bridged with mobile
phones. In recent years, the government has been trying to
improve access to government services by linking delivery to
mobile phones and mobile internet. With policy interventions
like Digital India (2015), the current government envisions a
digital society in which a cashless economy will feature
prominently, made possible by mobile banking and digital
payments. However, the challenges associated with imple-
menting technology-based policy interventions among a
diverse population have sparked a renewed debate on their
efficacy in addressing social problems (for example, Prakash
2016a; Mannathukkaren 2015).
The literature on the digital divide treats all ICTs as mono-
lithic. However, ICTs—like mobile phones—have various fea-
tures and functionalities (for example, voice calls, text mess-
ages, and internet applications) that all users may not operate
similarly (Selwyn 2004; van Dijk 2012). Furthermore, research
on the digital divide focuses predominantly on developed
countries (Reinartz 2016; Wei et al 2011). Given the increasing
reliance on ICTs to promote inclusive development in deve-
We express our sincere gratitude to the anonymous reviewer(s) for their loping countries, there is a need to explore whether differences
time and constructive feedback on the initial draft of the paper, as we exist in feature usage among mobile phone users in these
feel that they have helped produce a stronger manuscript. nations (Donner 2006). This paper adopts a feature-centric
Jang Bahadur Singh (jbs@iimtrichy.ac.in) teaches management approach to characterise the digital divide among Indian
information systems (MIS) at and M Vimalkumar (vimalkumar.f15005@ mobile phone users.
iimtrichy.ac.in) is a fellow at the Indian Institute of Management, Even though a number of government services are being
Tiruchirappalli with MIS as his subject area.
delivered digitally, particularly through mobile devices, to the
60 august 10, 2019 vol lIV no 32 EPW Economic & Political Weekly
SPECIAL ARTICLE

best of our knowledge, there is no countrywide study of India There are limited studies on the use of ICTs that address
to understand the scope of the digital divide based on users’ the characteristics or features of ICT devices. Scholars have
engagement with the various features of ICTs. Thus, the identified a range of “divides” among users in their approach to
objective of this study is to understand how users’ engagement different forms of ICT hardware. Pearce and Rice (2013) cate-
with various feature categories varies in different socio- gorise the research on disparities among users based on hard-
demographic groups. ware. For instance, based on internet usage patterns, Donner
et al (2011) propose four hardware-based access categories—
Understanding the Digital Divide neither computer- nor mobile-based access, computer-based
In the past decade, digital divide research has received signi- access, mobile phone-based access, and mobile and computer-
ficant attention from scholars in various disciplines—for based access—and show that there are differences among
example, social sciences, developmental studies, and information these various types of hardware users.
systems (Reinartz 2016). The literature identifies the nature These studies offer insights on the various kinds of digital
of the digital divide, its antecedents, and its consequences divides by exploring the differences in people’s use of various
(Wei et al 2011). features. The use of ICT features can be studied based on
Based on the literature review in Dewan and Riggins’ (2005) hardware or software (van Dijk 2005). While there are studies
study on the digital divide, Wei et al (2011) developed a on the uses of hardware features (for example, Chigona et al
framework with three levels to help understand the extant 2009; Donner et al 2011; Pearce and Rice 2013), to the best of
literature. In this framework, each successive level indicates a our knowledge, there have been no significant efforts to
higher order—they are digital access divide, digital capability understand differences in usage in terms of the software
divide, and digital outcome divide. The digital access divide features of ICT devices, particularly mobile phones.
focuses on inequalities in access to ICTs such as personal The software of an ICT device is essentially a combination of
computers, mobile phones, and internet in the home, school, multiple features that various users operate differently; thus,
or workplace (Dewan and Riggins 2005). The digital capabili- researchers have expressed the need for a feature-centric
ty divide refers to inequalities in ICT capabilities (Wei et al disaggregation of technology (for example, Donner 2006;
2011)—“the ability to use technology” when people have Jasperson et al 2005). Considering that the delivery of govern-
access to it (Dewan and Riggins 2005: 301). The digital access ment entitlements relies on mobile phone penetration and the
and capability divides, and other contextual factors, lead to range of mobile phone features available to users, it is useful
differences in outcomes caused by ICT usage, leading to a digital to disaggregate mobile phone features into voice calls, text
outcome divide (Wei et al 2011). The literature suggests that messages, and internet applications. Such a disaggregation
various key socio-demographic factors influence the digital will lead to a whole range of studies on the adoption, access,
access divide. Socio-economic status profoundly impacts the and outcomes of ICTs as tools for social inclusion.
access divide (Reinartz 2016; Hsieh et al 2011; Goldfarb and
Prince 2008). Age, in particular, can significantly influence Categorising Mobile Phones Based on Features
access to mobile phones and the internet (Chen 2013; Boase In this section, a brief history of mobile phone models is
2010; Rice and Katz 2003). Along with age, income and educa- presented and a framework to categorise the various features
tion are important factors that affect individuals’ digital skills of mobile phones is proposed. Mobile phone service started in
(Hargittai 2002). Studies show that people with higher educa- India in 1995, with the provision of basic features such as
tion levels and higher incomes exhibit a greater capacity to adopt making and receiving telephone calls and short message
information technology (IT) services (Akhter 2003; Lindsay service (SMS). First-generation mobile phones (also known as
2005; Bélanger and Carter 2009). Even though a few studies basic phones) are still popular in India due to their simplicity,
suggest that there is a disparity between men and women with low prices, and long battery lives (Agence France-Presse 2012).
respect to digital capability (Wei et al 2011; Sharma 2003), there As technology advanced, mobile phones incorporated more
is a lack of consensus on the role of gender in the digital divide features such as music players, cameras, and games. These
in studies carried out in countries like the United States and mid-range phones are also known as feature phones. They
Singapore (Reinartz 2016). were the first handsets to receive internet connectivity through
The existing research provides a rich understanding of the GPRS (General Packet Radio Services), popularly known as
digital divide; however, two issues remain under-explored. second generation (2G) internet technology (Singh 2012).
First, most studies in the digital divide domain conceptualise ICTs However, the mobile phone industry massively changed
as a monolithic black box (for example, Dewan and Riggins with the introduction of advanced computing platforms—
2005; Wei et al 2011; DiMaggio and Bonikowski 2008), instead popularly known as mobile operating systems—such as
of considering that users may operate some features but not Android, iOS, and Windows. These platforms supported mobile
others. Thus, an analysis of what features are accessible to users application development by third parties and created appli-
is needed. Second, the existing studies were predominantly cation delivery channels, such as the Play Store or App Store,
conducted in developed countries, and this data may not hold through which users could download and install various
true for the developing world. Thus, a feature-centric study of applications. These applications range from simple utility
the digital divide in the developing world is needed. applications like calculators and notepads to mobile games,
Economic & Political Weekly EPW august 10, 2019 vol lIV no 32 61
SPECIAL ARTICLE
Figure 1: A Framework to Disaggregate Mobile Phone Use Based on Feature Use wallets), and so on. Thus, such features appear at the higher
end of the complexity axis.
Complexity of the application

Group 2 Group 4 By arranging these features—according to complexity and


General features Advanced features internet dependency—on a 2*2 matrix, we obtain four cate-
High

eg, SMS, MMS, caller tunes eg, Mobile banking, e-commerce gories of mobile phone features. The basic features (Group 1)
rank low in complexity and do not require internet access. For
Group 1 Group 3
instance, making or receiving calls does not require linguistic
Basic feautres Social features
Low

skills nor internet access, so it is considered low in complexity.


eg, Making/receiving calls eg, Social media apps/browsing
General features (Group 2) encompass those that are high in
complexity, but which do not require internet access. For
No Yes
Dependency on the internet example, use of SMS necessitates language and typing skills.
Similarly, taking a picture with a camera phone needs the user
social media applications, health monitoring applications, and to adjust the lighting, focus, and zoom. These features do not
applications to access government services. Currently, both require internet access to function.
Android’s Play Store and Apple’s App Store host more than two Social features (Group 3) are generally user-friendly applica-
million applications each (Statista 2017). The phones that come tions that are comparatively low in complexity, but need internet
with these mobile operating systems are known as smart- access to operate. For instance, social networking applications
phones, and they are typically capable of internet connectivity. like Facebook, WhatsApp, and Twitter require internet access and
The features of these phones can be extended through the are relatively user-friendly, and users with fundamental language
myriad applications available through the app stores. skills, with minimal effort, can easily use these applications.
As explained above, mobile phones offer users various Finally, Group 4—advanced features—are complex and also
features and functionalities. These features can be grouped need internet access to function. For instance, e-commerce and
according to two broad characteristics—the complexity of the mobile banking applications require internet access, and users
feature and its dependency on the internet. Using these two must go through multiple steps to perform a single transaction.
characteristics, we propose a framework to categorise mobile Based on this framework, we analysed empirical data obtained
phone features into four distinct groups. Such a framework from a nationwide survey of Indian adults about their usage
facilitates empirical analysis by helping divide a range of of various mobile phone features. The details of the empirical
features into a finite set of categories. analysis and results are discussed in the next section.
The proposed 2*2 framework disaggregates mobile devices
based on their features (Figure 1). One important characteristic Research Methodology
of mobile devices in this era is internet dependency (van Dijk To understand whether disparities exist among individuals—
2012). We broadly categorise the applications available on with regard to their ability to use features in the four
mobile phones into two kinds—applications that require an categories—an empirical study was conducted and described
internet connection and applications that do not. For instance, the following sections.
activities like making phone calls and taking pictures do not
need access to the internet, whereas browsing through social Data collection and description: This study utilises an
media and mobile banking do. Hence, the features of mobile individual-level data set from InterMedia’s Financial Inclusion
phones can be grouped based on whether they require internet Insights Tracker Survey (InterMedia 2016). The data set features
access. In the framework, we conceptualise this characteristic information from an annual national representative survey
along the horizontal axis. with a sample size of 45,036 of Indian adults, aged 15 years
Along similar lines, the other characteristic of mobile appli- and above. The survey was conducted between 6 March and
cations considered in this framework is the complexity of the 10 April 2015, across 29 Indian states. Its breadth of coverage
application. Using the notion of medium-related skills, van is one of the data set’s strengths, as it represents people of
Deursen and van Dijk (2010) define complexity as the degree varying ages across states (Mukhopadhyay 2016). Primarily,
of effort and the set of skills required by users to access a the survey measures trends and market developments in digital
feature. The greater the degree of effort employed or the skills financial services, based on the first and second surveys
required to operate the feature, the greater the complexity, conducted in 2013 and 2014, respectively. It used a multistage,
and vice versa. For example, to make a phone call, users just stratified, clustered, and randomised sampling methodology,
need numeracy—they dial the number and press the call which generated a proportional distribution of samples across
button. Thus, such features are on the lower end of the all Indian states, based on the 2011 Census of India. The data
complexity scale. On the other hand, to make an e-commerce were collected through face-to-face interviews with individuals,
transaction on a mobile phone, users must be literate, be able each lasting an average of 49 minutes. The collected data set
to search for products, compare and choose products, add contained information on basic demographics, household
products to a virtual shopping cart, enter delivery and billing characteristics, and access to and use of mobile devices, mobile
addresses, decide whether to make the payment online (in money, formal financial ser vices, financial literacy, and other
which case they must have access to bank accounts or mobile general financial behaviour.
62 august 10, 2019 vol lIV no 32 EPW Economic & Political Weekly
SPECIAL ARTICLE

For our analysis, we have selected a part of the data set received SMS or MMS, opted for caller tunes, and took photos.
containing the variables of interest (see the next section for However, they did not perform any activities that required
details) and we recoded them to simplify the analysis. A internet access. Group 3—social users—includes those who
description of the recoded variables and a statistical summary made internet searches or used social media applications to
of the data set is provided in Table 1. The acquired data set gather information and interact with other users. Respondents
includes both, mobile phone users and non-users. This study who used Facebook, WhatsApp, or Twitter on their phones
only considers respondents with access to mobile phones. were categorised in Group 3. Finally, Group 4—advanced
Hence, the subsample included 37,672 respondents, of whom users—are those who used their phones for mobile banking, to
42.7% were male and 57.3% were female. download songs, and to perform other advanced operations.
The higher order groups also used the features used by the
Dependent variables: The data set obtained describes mobile lower order groups. Thus, Group 4 members use advanced
phone usage behaviour along with demographics and related features like mobile banking, along with some or all of the
information. The data includes information on the different features used by Groups 1, 2 and 3.
features that individuals use such as making and receiving
phone calls, sending text messages, browsing the internet, and Independent variables: Existing literature on the digital divide
using social networking sites. Respondents were asked whether notes that users’ abilities affect the digital divide on different
they had used any of these features in the recent past. Based levels. For instance, studies show that socio-demographic factors
on the respondents’ reported use of features, we categorised like age, income, education, and area of residence influence
them into four groups, in line with the proposed framework. the digital divide (Chen 2013; Hsieh et al 2011; Boase 2010;
These four feature-use categories are the dependent variables. Goldfarb and Prince 2008; Rice and Katz 2003). Similarly, we
Group 1—basic users—are those who either made or received believe that these variables might also affect the feature-use
calls in the past three months and did not perform any other divide. Hence, the first set of independent variables for this
mobile phone operations. Group 2—general users—are analysis includes demographic variables such as age, area of
those who, apart from making and receiving calls, also sent or residence, education, and occupation/work. Earlier studies on
Table 1: Descriptive Statistics and Variables gender have shown inconsistent out-
Variable Description Categories Number Average Min Max comes with regard to the effect of gender
Gender Gender Male 16,094 on the digital divide. While some studies
Female 21,578 show that men are more likely to adopt
Age Years 38 15 100
ICTs than women (Hoffman et al 2006;
Area Area of residence Rural 25,856
Meneses and Mominó 2010), others find
Urban 11,816
that these effects disappear once other
Education Highest level of Illiterate/no formal education 8,791
education Schooling until Class 8 14,289 variables like age or income are controlled
Classes 8 to 12 10,944 for (LaRose et al 2007; Rice and Katz
Diploma (no graduation) 817 2003). These mixed findings prompted
Graduate and above 2,831 us to consider the gender of respond-
Work Primary job (that is, the Full-time worker 6,246 ents as an important independent vari-
job where you spend Working part-time 6,086 able to study its impact on the feature-
most of your time) Self-employed 5,026
use divide.
Housewife 14,487
Mobile phone ownership—the number
Student 3,474
of mobiles in a household—affects
Not working 2,353
Ownership Owns mobile phone Yes 23,509
the computer self-efficacy of individuals
No 14,163 (Wei et al 2011). Studies show that own-
Num of Mob Number of mobile 1 0 10 ing mobile phones enhances individuals’
phones in household self-efficacy. In contrast, the number of
Need_Asst Needs assistance use Yes 11,906 phones in a household indicates access
the phone No 25,766
to phones and the availability of resources.
Eng_prof English proficiency None at all 19,724
Thus, we have considered access to mobile
Very bad 5,152
Somewhat bad 5,583
phones and resource availability as in-
Good 5,290 dependent variables. Given that most
Excellent 1,923 features on mobile phones use English
Bank_Ac Holds bank account Yes 25,857 (Prakash 2016b), users’ proficiency in the
No 11,815 language plays a major role in their use
Groups User groups based on Basic 23,182 of various applications. Similarly, users’
(dependent the mobile phone use General 11,205
ability to operate phones independently
variable) Social 2,736
Advanced 549 is a direct measure of their self-efficacy,
Source: Authors' calculation. and vice versa (Marakas et al 1998).
Economic & Political Weekly EPW august 10, 2019 vol lIV no 32 63
SPECIAL ARTICLE

Thus, we consider English proficiency and users’ ability to to test for the multicollinearity of categorical variables. There
operate phones independently as the other set of independent is no significant correlation among the variables.
variables in the feature-use divide analysis. Finally, another For the multinomial logistic regression analysis, we used
variable we have considered is bank account ownership. We the “mlogit” package available in the R software. The results
considered this variable because individuals with bank ac- of the regression (equation 1) are provided in Table 2. Since
counts are more motivated to use advanced features like mobile the multinomial logistic regression is functionally similar to
banking and e-commerce applications. These variables and the logistic regression, to interpret the coefficient we calcu-
the descriptive statistics are available in Table 1. lated the odds ratio. Considering the log-likelihood ratio
(ChiSquare value) and the significant p-value, the model is
Analysis method and model: The present study attempts to statistically acceptable. For a multinomial logistic regression,
predict the likelihood of users being categorised in the four McFadden’s Pseudo R2 of 0.16 is regarded as an acceptable fit
feature-use groups, based on their demographics and other (McFadden 1977).
characteristics. As the dependent variable has multiple cate- On analysing the data in the table, we find that gender is
gories, we use a multinomial logistic regression to analyse highly significant in predicting the likelihood of group mem-
the data. The multinomial logistic regression is an extension bership. Compared to women, men have significantly higher
of logistic regression and is employed when the dependent chances of being in Group 2 as compared to Group 1, that is,
variable is nominal and contains more than two categories. the odds ratio indicates that for every three men in Group 2,
This regression predicts the probability of a given member two women are likely to be in Group 2. The disparity is even
belonging to a category of the dependent variable (Stark- higher between men and women in Groups 3 and 4. Consider-
weather and Moske 2011). Independent variables can either be ing the odds ratio, for every four men in Group 3, there are two
dichotomous (that is, binary) or continuous (that is, interval or women. This clearly indicates that as the complexity of appli-
ratio in scale). To examine the feature-use divide, we concep- cations and the internet access requirement increases from
tualised the following model. Groups 1 to 4, the feature-use divide increases between male
Equation 1: Regression Model for Feature-use Divide and female users. The results indicate that women are more
α + β1 Gender + β2 Age + β3 Eng_prof + β4 Education likely to only use the voice call feature as compared to men,
when controlling for other factors. Similarly, the pattern
Groups = +β5 Ownership +β6 Work+β7 NumOfMob+β Need_asst
8 repeats for the rural and urban divide. The area of residence
+ β9 Area + β10 Bank_ Ac + ε variable is significant and the odds ratios are greater than one.
Before running the multinomial logistic regression model, we The results suggest that people living in urban areas have a
winsorised the age factor—the only ratio scale variable—to higher chance of being in Group 2 compared to people living in
three standard deviations to minimise the effect of outliers. rural areas. The likelihood of belonging to Groups 3 and 4 is
Appendix 1 (p 67) presents the matrix for Spearman’s correlation, even higher for urban as compared to rural residents, as the
Table 2: Output of the Multinomial Logistic Regression
Group 2 Group 3 Group 4
Parameters β Odds Ratio Pr(>|t|) p β Odds Ratio Pr(>|t|) p β Odds Ratio Pr(>|t|) p
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13)
(intercept) -1.084 0.338 0.000 *** -5.728 0.003 0.000 *** -7.965 0.000 0.000 ***
Gender 1 (male) 0.437 1.548 0.000 *** 0.852 2.345 0.000 *** 0.994 2.701 0.000 ***
Education2 -0.228 0.796 0.000 *** 0.254 1.289 0.161 -0.462 0.630 0.158
Education3 -0.065 0.937 0.123 0.973 2.645 0.000 *** -0.021 0.979 0.947
Education4 0.180 1.198 0.076 . 1.678 5.353 0.000 *** 0.857 2.355 0.017 *
Education5 0.329 1.390 0.000 *** 1.651 5.212 0.000 *** 0.951 2.588 0.005 **
Work2 0.022 1.022 0.608 -0.248 0.781 0.007 ** -0.327 0.721 0.067 .
Work3 0.471 1.602 0.000 *** -0.008 0.992 0.924 -0.317 0.729 0.056 .
Work4 0.353 1.424 0.000 *** -0.314 0.731 0.001 *** -0.400 0.670 0.024 *
Work5 0.411 1.508 0.000 *** 0.097 1.101 0.253 -0.041 0.960 0.778
Work6 0.070 1.073 0.235 -0.398 0.672 0.000 *** -0.570 0.566 0.003 **
NumOfMob 0.018 1.019 0.189 0.441 1.554 0.000 *** 0.457 1.579 0.000 ***
Need_Asst1 (Need _Asst = Yes) -0.024 0.976 0.360 0.196 1.217 0.000 *** 0.303 1.355 0.001 **
Bank_Ac1 (Bank_Ac = Yes) 0.237 1.268 0.000 *** 0.292 1.339 0.000 *** 0.453 1.574 0.000 ***
Eng_prof 0.119 1.126 0.000 *** 0.536 1.709 0.000 *** 0.658 1.930 0.000 ***
Age -0.016 0.984 0.000 *** -0.061 0.941 0.000 *** -0.052 0.949 0.000 ***
Area1 (urban) 0.125 1.134 0.000 *** 0.805 2.237 0.000 *** 1.186 3.274 0.000 ***
Ownership1 (ownership = Yes) 0.257 1.293 0.000 *** 1.773 5.890 0.000 *** 2.083 8.027 0.000 ***
Significance (p)codes:0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1.
Log-likelihood: -29,011.
McFadden R 2:0.15513.
Likelihood ratio test: chisquare = 10,655 (p-value = < 2.22e-16).
Source: Authors' calculation.

64 august 10, 2019 vol lIV no 32 EPW Economic & Political Weekly
SPECIAL ARTICLE

odds ratio is significantly greater than one. For every rural (Groups 2, 3 and 4), as compared to those who do not hold
resident in Group 3, there are two urban residents, and for bank accounts.
every rural resident in Group 4 there are three urban residents. The following sections discuss the implications on research
These results indicate the existing feature-use divide between and policy suggestions in light of our findings.
urban and rural residents—the divide is smallest in Group 1
and highest in Group 4. Discussion and Policy Implications
Consistent with existing studies (Niehaves and Plattfraut The Prime Minister, at the Digital India dinner, 2015, declared,
2014; Lam and Lee 2006), the analysis suggests that age is an “We have attacked poverty by using the power of networks
important variable in determining the extent of the digital and mobile phones to launch a new era of empowerment and
divide. With an increase in age, self-efficacy in the use of inclusion” (Press Information Bureau 2015a).
digital devices decreases. The results for age are reported in The Prime Minister’s words echo the strong supposition of
Table 1. The results indicate that a unit increase in age reduces the government that the digitisation of welfare services can
the chances of respondents belonging to Group 2 as compared improve their reach among the poor and marginalised, and,
to Group 1. Similarly, the results show that respondents’ particularly, that mobile phones and the internet can help to
likelihood of belonging to Groups 3 and 4 reduce even further elevate the quality of life of the marginalised. Thus, services,
with age. Thus, the feature-use divide increases with an such as e-government, financial inclusion, direct benefit tra-
increase in age differences among individuals. nsfer, and mobile money, are centred on digital technologies.
Contrary to our expectations, education and occupation do However, this study reveals that mobile phone usage is not
not significantly affect the feature-use divide. Although uniform, and, thus, the disparity among users can constrain
higher educational attainment was expected to significantly the government in achieving the desired outcomes. The study
reduce the feature-use divide, its significance is limited to a provides evidence of the differences in feature use between
p-value of 0.017. Occupation does not significantly affect the individual mobile phone users based on different socio-
feature-use divide, indicating the ubiquitous nature of demographics characteristics, such as gender, age, place of
mobile phones. residence, and education. Thus, to achieve more inclusive
Two interesting results emerged regarding proficiency in governance, the government needs a deeper understanding
English and mobile phone ownership, both of which affect the of the digital divide among mobile phone and internet users.
feature-use divide. Given that most applications on mobile Without such an understanding of the digital divide, the
phones, and the majority of content on the internet, are avail- efforts of the government to empower marginalised sections
able in English, proficiency in the language becomes impor- of society will not materialise as expected. This paper
tant. Hence, consistent with expectations, the results suggest attempts to enhance our understanding of the digital divide
that a person more proficient in English is more likely to be in in terms of feature use.
Groups 3 or 4, as compared to people with lower proficiencies. Mannathukkaren (2015) argues that development aimed at
Similarly, owning mobile phones might increase people’s self- the marginalised and impoverished by simply providing them
efficacy, as it may motivate them to try new and more the mobile and internet services to access government services
applications. The results indicate that people who own mobile is just “dilution,” as they have not caught up with technological
phones are significantly more likely to be in the advanced advancements. Observing the widespread presence of mobile
feature-use category (Group 4), as compared to people who do phones but not development, Prakash (2016a) claims that the
not own mobile phones. Thus, mobile phone ownership can features and facilities that the government offers in the form
assist in bridging the feature-use divide. of digital services are accessible only to certain sections of so-
Other variables considered in the study were the number of ciety. Anecdotal evidence shows that the demonetisation of
mobile phones in households and people’s dependency on 2016 did not affect many urban residents, especially since they
others to use mobile phones. We found that these variables tend to shop using credit/debit cards and use app-based taxi
were not significant predictors of membership in Group 2, but services. However, people residing in villages—who buy gro-
were highly significant for Groups 3 and 4, as compared to ceries from pushcarts, travel by autorickshaw, and use their
Group 1. This indicates that when there are more mobile phones just to make calls—suffered as a result of this monetary
phones in households and more family members using phones, upheaval (Prakash 2016b). The results of this study demon-
it leads to vicarious learning (Wei et al 2011) that helps the strate a disparity among users belonging to different feature-
focal people learn to operate more complex applications. Indi- use categories, based on their socio-demographic profiles. This
vidual independence in terms of operating mobile phones is following section focuses on the implications of our findings
significant, as people who can operate mobile phones for public policy design.
independently are more likely to be in Groups 3 or 4, as com- Gender inequality is a grave concern in India, which ranked
pared to those who need assistance to use phones. 125 out of 159 countries on the Gender Inequality Index (GII) in
Finally, the holding a bank account variable is also signi- 2015 (HDR 2016). This is apparent in the use of mobile phone
ficant and has an odds ratio greater than one. This indicates features. Women are adversely affected by the feature-use
that compared to Group 1, having a bank account increases the divide and primarily use basic features like voice calls. Thus,
chances of people belonging to advanced feature-use groups government policies should encourage or incentivise women to
Economic & Political Weekly EPW august 10, 2019 vol lIV no 32 65
SPECIAL ARTICLE

utilise more advanced applications. Anecdotal evidence shows Our analysis suggests that people who own mobile phones
that encouraging women to use ICTs can be advantageous, as utilise more features than those who borrow phones from
they provide women with opportunities to upgrade their others. Owning mobile phones allows people to experiment
learning and self-employment prospects (Viswanath 2017). with services and try new features.
Government initiatives such as the Pradhan Mantri Gramin Policy interventions targeting weaker sections of society,
Digital Saksharta Abhiyan (launched in 2017), which intends such as Digital India and the drive for cashless transactions,
to educate citizens to use digital technologies, must focus on rely on internet-dependent mobile phone features. This paper
educating women more effectively. reveals that most marginalised respondents tend to use mobile
Even though internet connectivity is rapidly expanding, driven phones only for voice calls (Tables 1 and 2). This means that
mainly by mobile data plans, the growth is seemingly inequita- the intended beneficiaries are the least capable of using the
ble between urban and rural regions. According to the Telecom features through which services are delivered, which might
Regulatory Authority of India (TRAI), by the end of 2016, 61.9% reinforce or even worsen inequality.
of people in urban India and only 13.7% of those in rural India While digitisation has its benefits, the use of mobile
had internet subscriptions (Krishnan 2017). The disparity phones to deliver government services must consider the
between rural and urban residents indicates a digital divide. socio -economic barriers that restrict ordinary citizens
Our findings reconfirm this feature-use divide. To make Digital from gaining access to these services. The major barriers to
India services accessible to rural residents, there must be a gaining access are the affordability of advanced technolo-
strong focus on building better network infrastructure in rural gies and the lack of digital education. Programmes like the
areas and making mobile internet connections affordable. National Digital Literacy Mission should incorporate feature
BharatNet, an ambitious project of Digital India, aimed at literacy to overcome the divide. A good example is the
establishing network infrastructure in rural India, envisaged decision to use voice-based interventions in the Jaankari
equipping one lakh gram panchayats with broadband connecti- (information) Project of Bihar, which allows illiterate
vity by March 2017 (BBNL 2016). However, as of March 2019, people to register right to information (RTI) applications via
43,179 gram panchayats have been connected, of which only mobile phones.
12,740 were operational (BBNL 2019).
Another significant factor affecting the digital divide is Conclusions
literacy, particularly English proficiency. Much of the content Considering the significance of mobile phones in developing
on the internet is in non-Indian languages. This results in the countries, this paper attempts to enhance our understanding
consumption of digital technologies being highly skewed to- of the digital divide. It demonstrates that people with access
wards those fluent in English. Rural residents who rely on to technology may not be able to use all its features equitably.
vernacular languages are limited to making voice calls and This study develops a framework to disaggregate the features
consuming audiovisual entertainment available on the inter- into four categories—basic, general, social, and advanced.
net. Though efforts are being made to translate applications Then, using large-scale survey data, the study reveals the
available in English to local languages, developers usually existing disparity among Indian mobile phone users—the
assume that applications specifically intended for marginal- feature-use divide. This is an exploratory study that utilises
ised users need not be appealing and user-friendly (Prakash secondary survey data that was originally collected for a
2016b). This results in low-quality and low-cost applications different purpose. The major limiting factor of the study
for rural residents. To overcome such disparities, developers was that the secondary data did not capture factors such as
should design applications with an understanding of the ca- income, which might impact the feature-use divide. Hence,
pabilities and needs of consumers. future research should consider a large-scale study focused
Finally, owning mobile phones is a major factor that can on the feature-use divide, which may uncover more explana-
reduce the feature-use divide across the different age groups. tory variables.
References Communication of the ACM, Vol 52, No 4, DiMaggio, P and B Bonikowski (2008): “Make
Agence France-Presse (2012): “Revealed! Difference pp 132–35. Money Surfing the Web? The Impact of Internet
between Mobile Phone, Feature Phone and Boase, J (2010): “The Consequences of Personal Use on the Earnings of US Workers,” American
Smartphone,” Hindustan Times, 26 July, http: Networks for Internet Use in Rural Areas,” Sociological Review, Vol 73, No 2, pp 227–50.
//www.hindustantimes.com/gadgets/revealed- American Behavioral Scientist, Vol 53, No 9, Donner, J (2006): “The Use of Mobile Phones by
difference-between-mobile-phone-feature- pp 1257–67. Microentrepreneurs in Kigali, Rwanda: Changes
phone-and-smartphone/story-g9oSjD3Lqce25d- to Social and Business Networks,” Information
Chen, W (2013): “The Implications of Social Capital
8MaU2BzI.html. Technologies & International Development,
for the Digital Divides in America,” The Infor- Vol 3, No 2, p 3.
Akhter, S (2003): “Digital Divide and Purchase
mation Society, Vol 29, No 1, pp 13–25.
Intention: Why Demographic Psychology Donner, J, S Gitau and G Marsden (2011): “Exploring
Matters,” Journal of Economic Psychology, Chigona, W, D Beukes, J Vally and M Tanner Mobile-Only Internet Use: Results of a Training
Vol 24, No 3, pp 321–27. (2009): “Can Mobile Internet Help Alleviate Study in Urban South Africa,” International
BBNL (2016): “BharatNet,” Vikashpedia, http://vi- Social Exclusion in Developing Countries?,” Journal of Communication, Vol 5, pp 574–97.
kaspedia.in/e-governance/digital-india/natio- The Electronic Journal of Information Systems Euromonitor (2017): “Mobile Phones in India:
nal-optical-fibre-network-nofn. in Developing Countries, Vol 36, No 7, pp 1–16. Country Report,” Euromonitor International,
— (2019): “Status of Bharatnet,” http://bbnl.nic. Dewan, S and F J Riggins (2005): “The Digital Di- https://www.portal.euromonitor.com/portal/
in/usage2.pdf. vide: Current and Future Research Directions,” analysis/tab.
Bélanger, Fand L Carter (2009): “The Impact of Journal of the Association for Infor mation Sys- Goldfarb, A and J Prince (2008): “Internet
the Digital Divide on E-Government Use,” tems, Vol 6, No 12, p 13. Adoption and Usage Patterns are Different: Im-

66 august 10, 2019 vol lIV no 32 EPW Economic & Political Weekly
SPECIAL ARTICLE
plications for the Digital Divide,” Infor mation Some Recent Developments,” Cowles Foun- Rice, R E and J E Katz (2003): “Comparing
Economics and Policy, Vol 20, No 1, dation Discussion, 474, New Haven, CT: Yale Internet and Mobile Phone Usage: Digital
pp 2–15. University. Divides of Usage, Adoption and Dropouts,”
Hargittai, E (2002): “Second-level Digital Divide: Meneses, Julio and Joseph Maria Mominó (2010): Telecommunications Policy, Vol 27, Nos 8–9,
Differences in People’s Online Skills,” First “Putting Digital Literacyin Practice: How pp 597–623.
Monday, Vol 7, No 4. Schools Contribute to Digital Inclusion in the Selwyn, N (2004): “Reconsidering Political and
HDR (2016): “Human Development Report: Human Network Society,” The Information Society, Popular Understandings of the Digital Divide,”
Development for Everyone,” document, UNDP, Vol 26, No 3, pp 197–208. New Media & Society, Vol 6, No 3, pp 341–62.
http://hdr.undp.org/sites/all/themes/hdr_ Mukhopadhyay, J P (2016): “Financial Inclusion in Sharma, U (2003): Women Empowerment through
theme/country-notes/IND.pdf. India: A Demand-Side Approach,” Economic & Information Technology, Delhi: Aut hors Press.
Hoffman, D L, T P Novak and A Schlosser (2006): Political Weekly, Vol 51, No 49, pp 46–54. Singh, Simran (2012): “Difference between Smart-
“The Evolution of the Digital Divide: How Neeraj, M (2016): “Mobile Internet Users in India phones and Feature Phones,” The Gadget Square,
Gaps in Internet Access May Impact Electronic 2016: 371 Mn by June, 76% Growth in 2015,” 12 November, http://the gadgetsquare.com/
Commerce,” Journal of Computer-mediated Dazeinfo, 8 February, https://dazeinfo.com/ 958/difference-between-smartphones-and-
Communication, Vol 5, No 3. 2016/02/08/mobile-internet-users-in-india- feature-phones/.
Hsieh, J J P-A, A Rai and M Keil (2011): “Addressing 2016-smartphone-adoption-2015/. Starkweather, Jon and Amanda Kay Moske (2011):
Digital Inequality for the Socioeconomically Niehaves, B and R Plattfaut (2014): “Internet “Multinomial Logistic Regression,” https://it.
Disadvantaged through Government Initia- Adoption by the Elderly: Employing IS Tech- unt.edu/sites/default/files/mlr_ jds_aug2011.
tives: Forms of Capital that Affect ICT Utili- nology Acceptance Theories for Understand- pdf.
sation,” Information Systems Research, Vol 22, ing the Age-related Digital Divide,” European
Statista (2017): “Number of Apps Available in
No 2, pp 233–53. Journal of Information System, Vol 23, No 6,
Leading App Stores 2017,” Statista, https://
InterMedia (2016): “The Financial Inclusion Insights pp 708–26.
www.statista.com/statistics/276623/number-
Program,” http://finclusion.org/. OECD (2001): Understanding the Digital Divide, of-apps-available-in-leading-app-stores/.
Jasperson, J S, P E Carter and R W Zmud (2005): “A Paris: OECD Publications, https://www.oecd.
Steinmueller, W E (2001): “ICTs and the Possi-
Comprehensive Conceptualisation of Post org/sti/1888451.pdf.
bilities for Leapfrogging by Developing Coun-
Adoptive Behaviours Associated with Infor- O’Riain, S (2004): The Politics of High Tech Growth:
tries,” International Labour Review, Vol 140,
mation Technology Enabled Work Sys- Developmental Network States in the Global
No 2, pp 193–210.
tems,” MIS Quarterly, Vol 29, No 3, pp 525–57. Economy, Cambridge: Cambridge University
Press. TRAI (2017): “Highlights of Telecom Subscription
Krishnan, Aarati (2017): “How Many Indians Have
Data as on 31 December, 2016,” TRAI Press
Internet?,” Hindu, 26 March, http://www.the Pearce, K E and R E Rice (2013): “Digital Divides
Release No 12/2017, http://www.trai.gov.in/
hindu.com/business/how-many-indians-have- from Access to Activities: Comparing Mobile
sites/default/files/Press_Release_11_17_Feb_
internet/article17668272.ece. and Personal Computer Internet Users,” Journal
2017_Eng_0.pdf.
Lam, J C Y and M K O Lee (2006): “Digital Inclu- of Communication, Vol 63, No 4, pp 721–44.
siveness—Longitudinal Study of Internet Prakash, Amit (2016a): “Enabling Human Capacity van Deursen, A J A M and J van Dijk (2010): “Inter-
Adoption by Older Adults,” Journal of Manage- in ICT Development Programmes,” Deccan net Skills and the Digital Divide,” New Media &
ment Information Systems, Vol 22, No 4, Herald, 9 July, http://www.deccanherald.com Society, Vol 13, No 6, pp 893–911.
pp 177–206. /content/556767/enabling-human-capacity- Van, Dijk, J A (2005): The Deepening Divide:
LaRose, R et al (2007): “Closing the Rural Broad- ict-development.html. Inequality in the Information Society, India:
band Gap: Promoting Adoption of the Internet — (2016b): “Design Logic in Digital, Less-cash Sage Publications.
in Rural America,” Telecommunications, Vol 31, Age Needs Reality Check,” Deccan Herald, — (2012): The Evolution of the Digital Divide: The
Nos 6 –7, pp 359–73. 12 December, http://www.deccanherald.com/ Digital Divide Turns to Inequality of Skills and
Lindsay, C D (2005): “Employability, Services for content/586119/design-logic-digital-less-cash. Usage, Amsterdam: IOS Press.
Unemployed Job Seekers and the Digital html. Viswanath, Vanita (2017): “A Digital Literacy Initia-
Divide,” Urban Studies, Vol 42, No 2, pp 325–39. Press Information Bureau (2015a): “Text of Speech tive Is Helping Improve the Lives of Women in
Mannathukkaren, Nissim (2015): “The Grand by Prime Minister at the Digital India Dinner Uttar Pradesh’s Villages,” Firstpost, 6 February,
Delusion of Digital India,” Hindu, 6 October, 26 September 2015, San Jose, California,” Gov- http://www.firstpost.com/living/a-digital-
http://www.thehindu.com/opinion/op-ed/ ernment of India, Prime Minister’s Office, literacy-initiative-is-helping-improve-the-
the-grand-delusion-of-digital-india/arti- http://pib.nic.in/newsite/PrintRelease. lives-of-women-in-uttar-pradeshs-villag-
cle7727159.ece. aspx?relid=128215. es-3362176.html.
Marakas, G M, Mun Y Yi and Richard D Johnson — (2015b): “Prime Minister Modi at Digital India Waverman, L, M Meschi and M Fuss (2005): “The
(1998): “The Multilevel and Multifaceted Char- Dinner: We Will Make Governance More Impact of Telecoms on Economic Growth in
acter of Computer Self-Efficacy: Toward Clari- Accountable and Transparent.” Developing Countries,” The Vodafone Policy
fication of the Construct and an Integrative Reinartz, A (2016): “Digital Inequality and the Use Paper Series, Vol 2, No 3, pp 10–24.
Framework for Research,” Infor mation Systems of Information Communication Technology,” Wei, K K, H H Teo, H C Chan and B C Tan (2011):
Research, Vol 9, No 2, pp 126–63. Doctoral Thesis, Universität Passau, https:// “Conceptualising and Testinga Social Cog-
McFadden, D (1977): “Quantitative Methods for opus4.kobv.de/opus4-uni-passau/frontdoor/ nitive Model of the Digital Divide,” Information
Analysing Travel Behaviour of Individuals: index/index/docId/365. Systems Research, Vol 22, No 1, pp 170–87.

Appendix 1: Spearman’s Correlation Table


1 2 3 4 5 6 7 8 9 10
1 Gender
2 Education -0.20
3 Work 0.37 0.09
4 NumOfMob -0.05 0.36 0.09
5 Ownership 0.30 -0.32 0.18 -0.35
6 Need_Asst 0.16 -0.21 0.05 -0.09 0.22
7 Bank_Ac 0.12 -0.15 0.11 -0.11 0.17 0.04
8 Eng_prof -0.15 0.70 0.12 0.36 -0.28 -0.18 -0.14
9 Age -0.07 -0.32 -0.21 -0.08 0.04 0.14 -0.19 -0.27
10 Area 0.01 0.22 0.07 0.26 -0.15 -0.06 -0.05 0.25 -0.01
11 Group -0.16 0.29 0.03 0.22 -0.20 -0.10 -0.09 0.32 -0.20 0.15

Economic & Political Weekly EPW august 10, 2019 vol lIV no 32 67

You might also like