Link&Kowal&Qureshi 2019

Information Systems Management
ISSN: 1058-0530 (Print) 1934-8703 (Online) Journal homepage: https://www.tandfonline.com/loi/uism20
Open Source in Development: Enabling Business

and Services
Georg J. P. Link, Jolanta Kowal & Sajda Qureshi
To cite this article: Georg J. P. Link, Jolanta Kowal & Sajda Qureshi (2019): Open Source
in Development: Enabling Business and Services, Information Systems Management, DOI:
10.1080/10580530.2020.1696548
To link to this article: https://doi.org/10.1080/10580530.2020.1696548
Published online: 27 Nov 2019.
Submit your article to this journal
Article views: 6
View related articles
View Crossmark data
Full Terms & Conditions of access and use can be found at

https://www.tandfonline.com/action/journalInformation?journalCode=uism20
INFORMATION SYSTEMS MANAGEMENT
https://doi.org/10.1080/10580530.2020.1696548
Open Source in Development: Enabling Business and Services

a b a
Georg J. P. Link , Jolanta Kowal , and Sajda Qureshi
a
Department of Information Systems and Quantitative Analysis, College of Information Science and Technology, University of Nebraska,
Omaha, Nebraska, USA; bDepartment of Historical and Pedagogical Sciences, Institute of Psychology, University of Wrocław, Wrocław, Poland
ABSTRACT KEYWORDS
This paper investigates the role of open source participation and employment in the service New business formation;
industry in development. We statistically analyze country-level data from publicly available global development outcomes;
databases. The findings suggest that open source participation and employment in the service ICT4D; open source
industry are together and individually positive moderators in the positive correlation of new software; employment in
service industry; innovation;
business formation and development outcomes. This paper contributes to socioeconomic devel- entrepreneurship; type of
opment by identifying ways in which open source participation contributes to development. economy
Introduction and cloud services (Lindman & Rajala, 2012). Open

source communities may offer an innovative model of
The use of information and communication technol-
peer production that enable new business exchanges,
ogies (ICT) in business has repeatedly been connected
thus leading to new forms of employment in the
to positive outcomes such as business growth,
services industry (Feller, Finnegan, Fitzgerald, &
increased productivity, administrative efficiencies,
Hayes, 2008; Fitzgerald, 2006). In this way, open
increased revenues, improved marketing strategies,
source software (OSS) can enable small and microen-
better access to customers, and cost saving (Qureshi,
terprises to spend their limited resources on hiring
Kamal, & Wolcott, 2009). With the advent of more
more employees instead of purchasing expensive soft-
widely available broadband internet, new forms of
ware and equipment. In some cases, the use of OSS
business strategies have emerged, and new innova-
can reduce a software’s cost of ownership (Wheeler,
tions for low resource environments have improved
2015). OSS can enable wealth to remain within the
the lives of people (Qureshi, 2010). Innovations and
local community by enabling skills for software devel-
development initiatives are most effective when rooted
opment and maintenance to be developed.
in local communities that know local needs and con-
Open source communities may create an environment
textual constraints (Nanne, Moshabela, Huynh, &
for collaboration and innovation (Garzarelli, Limam, &
Diop, 2015). Entrepreneurs innovate within their
Thomassen, 2008). Open source communities can be
communities, improve the lives of people, form new
welcoming to outside contributions and entrepreneurial
businesses, and are a source of development. While
engagement (The Mozilla Foundation & Open Tech
the exact connection between entrepreneurship and
Strategies, 2018). In most open source communities,
economic development may not be conclusive, this
innovations can be freely accessible and can be monetized
paper moves the field forward by adding to what is
by those that are able to adopt the software for business
known about the relationship between new business
purposes (von Hippel & von Krogh, 2003). The rights to
formation and development outcomes.
freely use, modify, and share open source software are
The rise of the service sector has been seen to be an
expressed in open source licenses approved by the Open
engine of growth in Western economies, when there is
Source Initiative.1 Innovations are made possible when an
an expansion of resources, a major resource discovery,
entrepreneur has the ability to code or hires someone to
or a significant productivity increase in services
code and contribute on their behalf. In such cases, benefits
(Swan, 1985). It has been suggested that open source
of this innovation model include access to license-cost
software provision may enable new services to be
free software; entrepreneurs can directly influence soft-
offered that increase productivity. Examples of such
ware development and ensure that desired and needed
services are computer support, infrastructure hosting,
CONTACT Georg J. P. Link glink@unomaha.edu Department of Information Systems and Quantitative Analysis, College of Information Science and
Technology, University of Nebraska at Omaha, Omaha, NE, USA
© 2019 Taylor & Francis
2 G. J. P. LINK ET AL.
features are implemented, thus improving the usefulness 2016). ICT and development have been subject to
in their local context (von Hippel, 2002). research for over thirty years (Walsham, 2017). The
Microenterprises are an important driver for devel- body of literature explores ICT and development in
opment (Grosh & Somolekae, 1996). Entrepreneurs a variety of ways. Brown and Grant (2010) identified
who run microenterprises often solve societal issues two streams of research: (1) the study of ICT for develop-
and create business models to support their activities. ment (ICT4D) with an interest in how ICT impacts devel-
Without the use of information and communication opment and (2) the study of ICT in development with an
technology, entrepreneurs tend to be disadvantaged interest of how people in low resource environments
and less effective in bringing about development engage with ICT. Additionally, Avgerou (2008) identified
(Duncombe & Heeks, 2002). As more cities and vil- three distinct assumptions that underlie current research
lages have internet access, more entrepreneurs can in ICT and development: (1) developing countries are
engage in open source communities and use open catching up to the technologically-advanced economics
source software innovations to spark business ideas, and theories of technology adoption dominate this dis-
reduce up-front sunk costs, lower entry barriers, and course; (2) ICT4D is about constructing new solutions for
gain visibility with customers (Gruber & Henkel, developing countries and focuses on local social context,
2006). Additionally, entrepreneurs find open source exploring local meanings, and striving toward locally
software to provide a suitable solution to support appropriate solutions; and (3) ICT4D reaches beyond
their business ideas and deliver services (Mitra, 2009). the local concerns to consider the implications of
In this paper, we investigate the role of open source dynamics resulting from the deployment of ICT on
participation and employment in the service industry in macro-level political and economic concerns. In the con-
development. We consider new businesses as the loca- text of this body of knowledge, our paper is an ICT for
tion where open source participation and employment development study – investigating the impact of new
in the service industry are instantiated and transformed business formation on development outcomes with open
into development outcomes. Our investigation is source participation and employment in service industry
important because governments make policies targeted as moderators on a country level – within the research
at increasing development and use fostering open stream that assumes that ICT4D is about constructing
source participation as one approach (Link & Qureshi, new solutions for developing countries. From this foun-
2018). We also consider that not all countries work the dation, we develop four hypotheses.
same, and therefore we test our findings across groups
of countries with different levels of development.
Development outcomes
The structure and flow of the argument in the fol-
lowing sections are as follows. The theoretical back- Development is “a process of expanding real freedoms
ground is reviewed in section 2. Sub-sections 2.1-2.6 that people enjoy” (Sen, 1999, p. 3). Development
develop hypotheses which are summarized and outcomes can be grouped into three dimensions: eco-
integrated into a theoretical model in section 2.7. nomic, social, and human (Malaquias, Malaquias, &
Section 3 details the methodology for investigating Hwang, 2017). Economic development can be
the theoretical model with details on the sample (3.1) observed in indicators such as economic growth,
and data sources (3.2). Section 4 presents the analysis growth in per capita income, registration of new busi-
and findings, discussing each hypothesis from the the- nesses, reduction in poverty, stimulated financial
oretical model in turn. Section 5 moves past the market liquidity, or unemployment rate. Social devel-
findings and discusses their implications for socioeco- opment can be observed in indicators such as reduced
nomic development (5.1) and businesses and govern- inequalities, social exclusion, (digital-) inclusion,
ments (5.2). Section 6 formulates conclusions and access to government services, crime levels, and level
highlights contributions. Section 7 discusses limita- of corruption. Human development is about enlarging
tions and paths to future research. human choices and can be observed in non-material
indicators related to a long and happy life, education,
and a decent standard of living (Malaquias et al.,
Theoretical background
2017). In this paper, we acknowledge that all develop-
We position our paper in the academic discourse as ment outcomes are connected.
follows. Information and communication technology The context of this paper is ICT for development
(ICT) impacts development by enabling government, with a macro-level perspective. One challenge in asses-
business, and personal activities (Roztocki & Weistroffer, sing development outcome is its many dimensions
INFORMATION SYSTEMS MANAGEMENT 3
(Kowal & Paliwoda-Pękosz, 2017). To find out if there Open source participation and business formation
are any correlations to development outcomes, we sur-
Open source software innovations can spark business
mise that development outcomes can be expressed in
ideas, reduce up-front sunk costs for development,
one construct – by combining several development
lower entry barriers, and gain visibility with potential
outcome indicators. This leads to our research assump-
customers (Gruber & Henkel, 2006). Successful new
tion, which is a prerequisite to our investigation:
businesses come from entrepreneurs that take advan-
Research Assumption RA: Development outcomes can be tage of opportunities by transforming knowledge
expressed in one construct by combining several devel- (Ardichvili, Cardozo, & Ray, 2003). An entrepreneur’s
opment outcome indicators. open source participation can provide the social net-
work and alertness that lead to an iterative learning
New business formation and development process (Argyris & Schön, 1978), and the development
outcomes of a knowledge corridor which leads to heightened
alertness to new opportunities. As such, open source
The formation of new businesses is traditionally seen participation is a form of entrepreneurial activity that
as an indicator of economic development. Specifically, may lead to the creation of new businesses.
micro- and small enterprises (MSE) are the founda- Entrepreneurs can participate in open source com-
tion of economic development (Grosh & Somolekae, munities in different ways (Dahlander & Magnusson,
1996). MSE without ICT rely on localized, informal 2005; Lindman & Rajala, 2012). Lurking entrepreneurs
social networks for information and knowledge which are lightly engaged through watching open source com-
are often of poor quality and thus limit MSE’s influ- munity activities for new ideas and using software with-
ence on social and economic development (Duncombe out having to pay license fees. Interacting entrepreneurs
& Heeks, 2002). ICT has been reported to cause posi- are more engaged through seeking help, posting ques-
tive outcomes such as business growth, increased pro- tions, reporting bugs, and requesting features. These
ductivity, administrative efficiencies, increased are also available with commercial software programs
revenues, improved marketing strategies, better access that often involve installation of beta versions of soft-
to customers, and cost savings (Bharati & Chaudhury, ware where the entrepreneur can give feedback in
2009; Qureshi et al., 2009). Furthermore, the use of exchange for a reduced price. Contributing entrepre-
ICT in MSE has been known to increase their growth neurs have the opportunity to be fully engaged by
by a factor of 3.8% (Qiang, Clarke, & Halewood, submitting software patches, writing documentation,
2006). updating wiki entries, helping others, discussing design
However, several barriers to ICT use in MSE exist. decisions, defining standards, and facilitating commu-
For some time, existing MSE did not see the benefit of nity activities. The fully engaged participation has the
bringing ICT into their firm and only adopted ICT most potential for learning and benefitting from open
because of social pressure to do so (Riemenschneider, source (Bonaccorsi, Giannangeli, & Rossi, 2006; Nagle,
Harrison, & Mykytyn, 2003). MSEs that want to adopt 2018). The choice of participation may be shaped by
ICT might lack technical skills or cannot afford the the governance of an open source community and its
investment in technology (Qureshi et al., 2009). This willingness to accept outside contributions (Di Tullio &
problem may be exacerbated when adopting OSS Staples, 2013; The Mozilla Foundation, & Open Tech
instead of more user-friendly, polished commercial Strategies, 2018).
software. Nevertheless, many entrepreneurs are seeking Technical competencies and coding skills are an
to upgrade their ICT because they see the benefits of important prerequisite to contributions. If the entre-
being better connected with customers and having preneur is not a programmer, they cannot create and
access to timely information (Donner, 2006). For this submit patches, add-ons, and modifications. In such
reason, we surmise that in today’s age of information cases, entrepreneurs can pay employees with the requi-
technology, new businesses are even more so drivers of site coding skills to participate in open source commu-
development outcomes. This leads to the formulation nities and to pursue the goals of the entrepreneur
of our first hypothesis, which is the baseline for our (Dahlander & Wallin, 2006). As long as the entrepre-
investigation: neurs are programmers or willing to pay for them,
organizational innovation processes have been shown
Hypothesis H1: New business formation and develop- to benefit from open source participation as entrepre-
ment outcomes are correlated. neurs learn from other open source participants and
help shape open source software to meet business needs growth of traditional manufacturing and logistics based
(Germonprez et al., 2016). Such benefits entail effort industries. Chapman, Soosay, and Kandampully (2002)
and expense that may not be readily available for a new suggested that these independent companies undertak-
business. In which case they may create their own ing service-based activities in those sectors are defined as
homegrown software depending upon what the “service industries.” In most countries, the service sector
requirements may be. is seen to bring about employment increases, and there-
Open source communities have been criticized for fore growth. This occurs when there is an expansion of
ICTs that may be implemented with intentions that are resources, or major resource discovery and significant
not aligned with development goals (Sahay, 2016). productivity increases in services (Swan, 1985).
While ICT is often framed as a silver bullet for devel- Employment in the services sector also supports growth
opment or used synonymously with development, it is in other sectors (Wegner, 1987). In particular, the ser-
the livelihood of people and their neighborhoods that vices sector powers economic development in many
matters. Both criticisms can be responded to by invol- industrialized countries (Breitenfellner & Hildebrandt,
ving local community members in the development 2006). This is because new businesses are easier to
projects and respecting their specific needs and local form in the services sector. In countries that have higher
knowledge (Nanne et al., 2015). Open source commu- levels of skilled labor and education, the portion of small
nities provide such a space where members can con- and microenterprises in the services sector tends to be
tribute to the software development and innovation higher (Grosh & Somolekae, 1996).
occurs quickly to address real problems and needs. Increased competition in the global marketplace has
Entrepreneurs benefit from this engagement through compelled every industry to transform itself into a truly
the learning of real-world issues and available solutions customer-oriented, service-focused enterprise, irrespec-
that can subsequently inform their entrepreneurial tive of the products and services it sells (Chapman
activities (Lindman & Rajala, 2012). et al., 2002). Countless service-oriented businesses are
When entrepreneurs make decisions that benefit now assisting traditional corporations with the acquisi-
their microenterprise and their local community, they tion and retention of their customers. As ICTs enable
create shared value (Porter & Kramer, 2011). Porter service firms to improve their efficiency, effectiveness,
and Kramer (2011) offer a perspective on creating and level of service, the service sector has been one of
shared value, which occurs through policies and the fastest adopters of ICTs and has fueled employment
operating practices that enhance the competitiveness in the service sector. Chapman et al. (2002) contend
of a microenterprise while simultaneously advancing that ICTs have been a major driver for growth and
the economic and social conditions in the local com- innovation in the services sector. This leads us to our
munity. It appears that shared value can be created third hypothesis:
through participation in open source communities
where entrepreneurs create value for themselves and Hypothesis H3: Employment in the service industry is
the community they share their innovations with. a moderator that affects the correlation between new
Research shows that this approach can raise the well- business formation and development outcomes.
being of entire communities (Leong, Pan, Newell, &
Cui, 2016). Participation in open source communities
Open source software and employment in the
follows mechanisms by which entrepreneurs benefit
services industry
themselves and others (Dahlander & Magnusson,
2005). This leads to the formulation of our second Open source communities offer an innovative model of
hypothesis: peer production that enable business exchanges in open
source service networks, thus leading to new forms of
Hypothesis H2: Open source participation is a moderator employment in services provision (Feller et al., 2008;
that affects the correlation between new business forma- Fitzgerald, 2006). The use of OSS by small and micro-
tion and development outcomes. enterprises can enable growth in their business, if the
growth is measured in terms of hiring more employees
to be able to run and maintain the software. For many
Employment in the service industry
microenterprises, the functionalities offered by com-
The service sector is seen as an engine of growth strong mercial software is often outside their reach due to
enough to convey substantial benefits to Western econo- their limited resources. Some authors suggest that the
mies (Swan, 1985). The rapid growth of the service total cost of ownership for open source software is
sector is attributable to the ecosystems surrounding the often competitive even when accounting for hiring
someone to manage and maintain it (Wheeler, 2015). have reached enterprise-grade, which makes it a viable
Instead of paying license fees, hiring a local expert building block for successful businesses (Feller et al.,
retains created wealth within the local community and 2008; Fitzgerald, 2006; Wheeler, 2015). Entrepreneurs
builds capacity for development (Jacovkis, Komen, & can use open source software to build on existing solu-
Tucker, 2010). tions, customize or extend them, and create unique
Contrary to expectations, the savings from some OSS solutions for their customers (Gerber, 2016). Users of
implementations can be lost by hiring specialists to deal open source software can recommend changes, report
with installing and maintaining the OSS. OSS implemen- bugs, and help improve the software, which entrepre-
tations are by no means equally beneficial to all busi- neurs benefit from without having to pay employees to
nesses. They have to be analyzed on a case-by-case basis. do all the work (Bhatt, Ahmad, & Roomi, 2016; von
The advantages of OSS may not always be forthcoming Hippel, 2002). The only difference with commercial,
as some OSS may be in its beta or rudimentary stages, closed source, is that the entrepreneur cannot modify
harder to install and maintain than commercial solu- the software themselves. Overall, an entrepreneur can
tions. This means that specialists have to be hired for build a service business with the help of open source
jobs that users would normally be able to do with com- software and dynamically respond to market changes
mercial software. This means potential savings in licen- and benefit from shared software innovations.
sing costs but potential costs for specialists. When entrepreneurs solve customers’ needs by
The increasing access to broadband internet and the modifying open source software, they can contribute
adoption of ICT increase the chance that entrepreneurs those changes back to the open source community and
start new businesses with ICT support from the incep- allow benefit to the members (Bhatt et al., 2016). Some
tion. For example, a local farmer and entrepreneur open source software licenses contain provisions that
provided services to other farmers and helped them make the sharing of modified open source software
be more productive through soil sampling and giving mandatory, which further ensures that all innovations
farming-related advice (Jha, Pinsonneault, & Dubé, are made available. Some companies avoid OSS made
2016). The entrepreneur profited from providing the available under those licenses because they could build
service, and the customers increased their returns from on it, but then they would have to give back the added
harvests and became wealthier. By responding to the parts. These could be used by the competitors for free
new and changing needs of their clients, the entrepre- while the costs of programmers and equipment were
neur provided new services and through the ongoing incurred by the contributing company.
innovation continued to create shared value. The soft- Open source software can support rapid economic
ware tools were available for other entrepreneurs to growth in low resource environments because the
replicate this service in their communities. The entre- development model allows for a division of labor,
preneurs were part of a network that collectively inno- increases collective and individual knowledge, fosters
vated new services. sharing, and is scalable (Garzarelli et al., 2008).
Several characteristics of open source software make According to Fitzgerald (2006), the open source phe-
it amenable to support services: open source software is nomenon represents a radical change in the software
a technology that allows entrepreneurs to innovate very landscape leading to fundamental changes in the devel-
quickly, use existing solutions, and build an agile and opment process, reward mechanisms, distribution of
scalable business (Castelluccio, 2008). Open source work and the business models that govern profit-
software is marked by license terms that allow anyone making. Services offered include computer support,
to use the software for any purpose, share the software infrastructure hosting, and other ICT services that can
freely, make changes to the software, and share those be built from open source software. This leads us to the
changes while still allowing businesses to make money following hypothesis:
with it (Kelty, 2008).
The open source software licensing models may Hypothesis H4: The two moderators, open source parti-
enable monopolies to be avoided by offering greater cipation and employment in the services industry, have
access to software innovations (von Hippel & von a stronger influence together than each individually.
Krogh, 2003). There are many different types of open
source software licensing models: some require that
code under one license is not allowed to be mixed
Countries are different
with code under another – and some licenses even
explicitly prevent the use of the software for profit. On the basis of literature (Kowal & Paliwoda-Pękosz,
The quality and stability of some open source software 2017; Kowal & Roztocki, 2013), and our initial analysis,
we assume that countries with different levels of devel- enable development outcomes to be achieved. This
opment should be considered separately. Economic model shows that new business formation (NBF) and
data suggests that a linear correlation between new development outcomes are correlated (H1). We expect
business formation and development outcomes may new business formation to have a positive effect on
not be the best model (Wennekers, van Stel, Thurik, development outcomes and vice versa. Since ICTs
& Reynolds, 2005). Different levels of development enable activities of entrepreneurs, this model suggests
have been linked to different stages of economic devel- that open source participation (GIT) is a moderator
opment which provide varying incentives and opportu- that affects the correlation between new business for-
nities for entrepreneurs (Acs, Desai, & Hessels, 2008). mation and development outcomes (H2). The variable
Low resource environments have lower standards of GIT is derived from GitHub, the most popular open
living, weak industrial and commercial base, and poor source participation platform. The services industry
infrastructure, thus making entrepreneurship some- and especially services provided by microentrepreneurs
times the only means to earn a living. In contrast, tend to power new business formation and develop-
advanced–developed countries have a high level of ment outcomes. We thus posit that employment in the
gross domestic product (GDP) per capita, as well as service industry (SER) is a moderator that affects the
a very significant degree of industrialization, commer- correlation between new business formation and devel-
cial base, high standards of living, and a well-developed opment outcomes (H3). When microentrepreneurs use
infrastructure (Roztocki & Weistroffer, 2016). Our the- open source software, their ability to form new busi-
oretical model might work differently for countries of nesses and offer successful development outcomes
different levels of development. This leads to our testa- increases. The model illustrates this by showing that
ble statement, where we evaluate our hypotheses across two moderators – open source participation and
different countries: employment in the services industry – have a stronger
influence than one individually (H4). Since some emer-
Testable Statement TS: The correlations of hypotheses
ging countries have low standards of living, a weak
H1-H4 vary by group of countries.
industrial and commercial base, and poor infrastruc-
tures compared to more advanced countries, these cor-
relations will vary by group of countries (TS).
Theoretical model
The following graphical depicture of our theoretical
model (Figure 1) illustrates the relationships between
Methodology
the four hypotheses and development outcomes. This
model assumes that development outcomes (DEV) can This research follows a quantitative deductive approach
be expressed in one construct by combining several to test the four hypotheses developed in the previous
development outcome indicators (AR). This also section. We used different statistical methods depend-
enables us to identify socioeconomic factors that may ing on the requirements of each hypothesis. A detailed
Figure 1. Theoretical model showing the four hypotheses in relation to each other. H1-H4 – research hypotheses, RA – research
assumption, TS – testable statement.
description of the statistical methods is provided inter- scales, including nominal, ordinal, interval, and ratio
leaved with the analysis. scales. Transformation was performed to compare
results and graph profiles on the same scale. First, we
standardized indicators to the z-score scale and then to
Sample
the T scale (Neukrug & Fawcett, 2014). The value of
Our investigation is at the global level. Each country z-score is the distance between the raw score and the
forms a data point. Some countries had to be excluded population mean in units of the standard deviation.
due to lack of data. We had sufficient data from 141 The z-score has a mean of 0, a standard deviation of
countries for the complete analysis. 1, and typical results between −1 (below the mean) and
We assumed that development outcomes based on 1 (above the mean) (Larsen & Marx, 1986). Second, we
investment do not actualize immediately but lag in calculated the T-score which is easier to understand
time. We further assume that any investment should because it has a mean of 50 and a standard deviation
not be measured as a point in time but as an effort over of 10. Thus, on a T scale, the average score is between
a longer period. We chose an investment period of 40 and 60. Appendix 1 also includes the abbreviations
three years (2011–2013) in which we calculated an for each variable.
aggregate of our independent variables. To give enough
time for investments to impact development outcomes,
we considered delayed dependent variables two years
Analysis and findings
later (2015). These years were chosen based on the
most recent available data available for the largest num- Construct: development outcome
ber of countries.
Research Assumption RA: Development outcomes can be
Our moderator variable (GIT) is open source parti-
expressed in one construct by combining several devel-
cipation on GitHub. For data, we turn to GitHub
opment outcome indicators.
because it is the most widely used open source colla-
We assume that Development Outcomes (DEV) can
boration platform and an established data source in
be one construct by combining several development
open source research (Cosentino, Luis, & Cabot,
outcome indicators. The purpose is to have one global
2016). In measuring open source participation,
DEV construct to use in our further analysis. To con-
GitHub introduces a bias because it captures only
struct our DEV construct, we consulted theoretical
a fraction of all possible open source participation for
foundations and definitions of socioeconomic develop-
two reasons. First, other platforms exist for open source
ment put forth by Kowal and Roztocki (2013), Stec,
participation. GitHub is a collaboration platform for
Filip, Grzebyk, and Pierscieniak (2014), Roztocki and
software development and as such hosts more than
Weistroffer (2016), and Kowal and Paliwoda-Pękosz
85 million2 software repositories, compared to half
(2017). Socioeconomic development is a process of
a million on SourceForge,3 the second largest platform.
changes or improvements in social and economic con-
This makes GitHub the best option for data to estimate
ditions related to an individual, an organization, or
the level of open source participation. Second, entre-
a whole country. On an individual level, socioeco-
preneurs can participate in many different ways (lurk-
nomic development manifests positive changes in
ing, interacting, and contributing) but only if they
socioeconomic status, manifested by among many
engage through GitHub will we be able to measure it.
indicators (Kowal & Paliwoda-Pękosz, 2017; Kowal &
For example, GitHub does not include mailing list or
Roztocki, 2013; Roztocki & Weistroffer, 2016),
chat capabilities which are widely used collaboration
including:
modes in open source. By focusing our analysis on
open source participation that is evident in contribu-
tions to the software, we eliminate concerns of counting – personal income, depending on economic growth
interactions that have less potential for an iterative (GNI per capita);
learning process and the development of a knowledge – personal wealth, which is included in different
corridor that leads to heightened alertness to new dimensions of Human Development Index
opportunities. (HDI) (Yakunina & Bychkov, 2015);
– level of education, which also manifests by mean
years of schooling (MSCH);
Data sources
– occupation, quality of life, standard of living, and
We used public data sources (see Appendix I for source general health, which lead to longer life expec-
information) which employed different measurement tancy (LE); and
– innovations from the perspective of human capital – improved labor market through competitiveness
(HC) and information and communication tech- and its global index (GCI); and
nologies (ICT), which mean the capability of – innovations in global scale that can be measured
developing novel competences such as new ICT by the Global Innovation Index (GII), which pro-
knowledge, skills, social and managerial compe- vides detailed metrics about the innovation per-
tencies at the workplace (Kowal & Jasińska- formance of countries and economies around the
Biliczak, 2016). world (Kowal & Paliwoda-Pękosz, 2017).
At the organizational level, socioeconomic development We built our new construct DEV based on the above
may be manifested: dimensions to indicate levels of socioeconomic devel-
opment and innovativeness. Specifically, DEV
– improvements in global competitiveness, which includes development of ICT organizational and tech-
can be measured in global by index GCI; nical infrastructure (IDI), human development (HDI),
– organizational income, which can be measured on life expectancy (LE), mean years of schooling
the country level by economic growth (GNI per (MSCH), economic growth (GNI per capita), level of
capita); innovativeness (II Score), innovativeness efficiency
– consumer demands, which require competitiveness (II Efficiency), human capital development (HC) and
that can be expressed by an index of global competi- global competitiveness (GCI). Because these serve as
tiveness (GCI) which is a factor of economic growth our dependent data, we collected their values for
that forces innovation and increases productivity the year 2015 from the data sources listed in
(Stiglitz, 2002; Stiglitz, Sen, & Fitoussi, 2010); Appendix 1. We used structural equation model
– business reputation and brands, and the quality of (SEM) and confirmatory factor analysis (CFA)
the workforce which can be expressed by human (Table 1, Figure 2) (Bagozzi & Yi, 2012; Davidov,
capital index (HCI) (Kowal & Paliwoda-Pękosz, 2008; Thompson, 2004) to test our research assump-
2017; Roztocki & Weistroffer, 2016); and tion RA that development outcomes can be expressed
– generally by innovations that support economic in one construct by combining several development
growth, which means the capability to introduce outcome indicators.
new or modernized products, technology or In Figure 2, our construct DEV is the measurement
enhanced organizational and technical processes model relating measured observable variables (within
(Kowal & Jasińska-Biliczak, 2016; Kowal & rectangles: for example [IDI], [HDI], [LE]) to latent,
Paliwoda-Pękosz, 2017; OECD/Eurostat, 2005, directly unobservable variable as Development
p. 49; Yunis, El-Kassar, & Tarhini, 2017); Outcomes (DEV). We assumed that DEV manifests
– special role of ICT innovations, which can be by concrete values of indicators. Variables d1,…, d10
expressed in global by the ICT Development (within ellipses) are residuals, thus represent the influ-
Index (IDI) (ITU, 2017). ence of factors not included in this model.
– increased national product and wealth, visible in In order to test the reliability and relevance of our
economic growth (GNI per capita); new construct DEV, we used CFA. For the CFA, we
built a model (i.e., model of DEV) in which we assume
At the country level, socioeconomic development is a certain set of factors (our indicators in rectangles) and
reflected by: by analyzing the values of random variables. We
Table 1. The Structured Regression Model (SEM) for Development Outcomes (DEV).
Construct and model Equation
(Factor)–>[dependent variable] R in the model λi Err. λ: di t p< Model data fit
The best global model 1 ∑dep.var = λ1*DEV+d1 0.98 0.01 0.001 AVE 0.86
(DEV)–>[IDI] 0.95 IDI = λ11*DEV+d11 1.00 0.00 189.03 0.001 Α 0.8
(DEV)–>[HDI] 0.96 HDI = λ12*DEV+d12 0.96 0.01 0.001 R 0.6
(DEV)–>[LE] 0.88 LE = λ13*DEV+d13 0.98 0.01 87.89 0.001 χ2/df 0,5
(DEV)–>[MSCH] 0.89 MSCH = λ14*DEV+d14 0.79 0.04 138.34 0.001 RMSEA < 0.01
(DEV)–>[GNI] 0.78 GNI = λ15*DEV+d16 0.95 0.02 17.75 0.001 p< 0.5
(DEV)–>[II Score] 0.92 II Score = λ16*DEV+d17 0.59 0.08 64.28 0.001 GFI 0.98
(DEV)–>[II Efficiency] 0.65 II Efficiency = λ17*DEV+d18 0.95 0.01 7.44 0.001 AGFI 0.90
(DEV)–>[HC] 0.92 HC = λ18*DEV+d11 0.50 0.08 79.42 0.001
(DEV)–>[GCI] 0.62 GCI = λ19*DEV+d11 0.98 0.01 6.35 0.0001
Symbols: AVE – average variance extracted, α – Cronbach’s α, λ – factor loading, Err. λ – – standard error of factor loading, t – Students’ t statistics, r – mean
correlation between positions, R – correlation between DEV and indicator (position), p – observed probability, χ2/df, RMSEA, GFI – indicators of model fit.
Figure 2. CFA Model of Development Outcomes DEV to test research assumption RA.
examine the validity of our assumption and estimate New business formation and development
the parameters of the model (e.g., factor loadings λ). outcomes
The method of assigning a normalized t or z score
to the DEV variable and its components address the
Hypothesis H1: New business formation and develop-
issue of effect size. For example, Table 1 presents the
ment outcomes are correlated.
values of factor loadings λ for particular dependent
variables, denoting DEV components (columns
We surmise that Development Outcomes (DEV) can be
assigned as Equations and λ). The values of λ mean
correlated with New Business Formation (NBF). We
the size effect or how a given ingredient will grow, i.e.,
hypothesized that DEV could be predicted on the basis of
how many tannins can increase or decrease GNI, if
concrete values of NBF. Thus, a linear regression was
DEV increases by one tannin. A sample equation:
calculated to predict DEV based on NBF. As can be seen
GNI = λ15 * DEV + d16, for λ15 = 0.95 and d15
from Table 2 (Hypothesis 1), a significant regression equa-
= 0.02 explains that if DEV increases by one tannin,
tion was found (F(1, 103) = 22.23, p < .01), with an adjusted
GNI will increase by 0.95 tannin, which means
R squared of 0.17. The regression coefficient R = 0.42
a specific growth effect. If we wanted to know the
indicated medium strength of dependence between criter-
specific effect value in the form of a raw result (source
ion variable DEV and the independent variable NBF. Thus,
data), we can transform from a normalized result of
we can predict DEV as equal to DEV ¼ 0:40 NBF. This
t or z to source values, which is a trivial task according
equation shows that DEV increased 0.40 standard devia-
to the formulas provided by Ferguson and Takane
tions for each standard deviation increase of NBF. At the
(1989) (see Tables 1 and 2).
global level, NBF was a significant predictor of DEV. Thus,
The quality of the CFA model is determined,
the hypothesis H1 is supported.
inter alia, by examining the values of the factor
loadings (λ) and the percentage of the variance
explained (AVE), along with the model’s fit for the
Investigating moderators
data (RMSEA, GFI, AGFI, χ2/df). In the affirmation
model (based on the logic of structural equation The typical situation in which moderators are sought is
analysis), the value of the charge is the value of the the presence of weak or unreliable relationships that
path factor corresponding to the path connecting sometimes occur and sometimes disappear, such as the
the given sub-index with the latent variable (latent) influence of NBF on DEV. The identification of mod-
(Bagozzi & Yi, 2012; Davidov, 2008). As can be seen erators is therefore of great practical importance, as it
in Table 1, the global construct DEV is characterized allows for determining the conditions in which
by very good data fit with indicators supporting the a relationship occurs and distinguishing them from
discriminatory power, validity, and reliability the conditions in which the relationship disappears,
(Bagozzi & Yi, 2012; Davidov, 2008). even if we do not understand why this is happening
The results of the empirical correlations and the (Baron & Kenny, 1986; McClelland & Judd, 1993). A
SEM analysis (see Table 1) support our research visual comparison of our variable means across groups
assumption RA, which gives us the confidence to use of countries is shown in Figure 4 to provide an impres-
the DEV in our subsequent analysis. sion of the data before we begin our analysis.
Table 2. Verified models of moderation in relation to research hypotheses.

Gr2: Gr3: Gr4: Gr5: Gr6:
Gr1: Emerging and Latin America Commonwealth of Emerging and Middle East. North Gr7:
Advanced Developing and the Independent Developing Africa. and Sub-Saharan
Group All (Global) economies Europe Caribbean States Asia Pakistan Africa
Hypothesis 1 DEV ¼ b1 NBF þ a
R 0.44*** 0.15 0.5* 0.39 0.54* 0.06 0.84** 0.65**
R Squared 0.18 0.02 0.24 0.15 0.29 0.00 0.70 0.42
Adjusted 0.17 0.01 0.14 0.09 0.14 0.00 0.65 0.37
R Squared
F(df1,df2) 24.99(1,10) 0.66(1,8) 2.29(1,7) 2.47(1,14) 2(1,5) 0.03(1,8) 11.94(1,5) 9.27(1,13)
p 0.0001 0.423 0.174 0.138 0.216 0.867 0.018 0.009
b1 0.40*** −0.14 0.50* 0.39* 0.54* 0.06 0.84** 0.65**
pr1 0.44*** −0.15 0.50* 0.39* 0.54* 0.06 0.84** 0.65**
r1 0.44*** −0.16 0.5* 0.39* 0.54* 0.06 0.84** 0.65**
Group All Gr1 Gr2 Gr3 Gr4 Gr5 Gr6 Gr7
Hypothesis 2 DEV ¼ b1 NBF þ b2 GIT þ b3 NBF GIT þ a
Multiple R 0.62*** 0.84* 0.74* 0.41 0.79 0.56 0.80* 0.62*
R Squared 0.38 0.70* 0.55* 0.17 0.63 0.32 0.64 0.38
Adjusted 0.36 0.66* 0.39* 0.04 0.26 0.22 0.57* 0.33*
R Squared
F(df1,df2) 19.41(3,99) 18.49(3,23) 3.61(3,6) 1.3(3.13) 1.7(3,4) 3.25(3,6) 8.92(3,4) 8.02(3,12)
p 0.0001 0.000 0.085 0.316 0.304 0.102 0.030 0.003
b1 1.70*** 5.31* 0.20 0.35 4.11 0.05 0.84** 0.29
b2 0.58*** 1.25* 0.53* 0.13 1.32 0.42 −0.05 0.21
b3 −1.35*** −5.38* 0.42 0.36 −4.25 0.51 0.84** 0.59**
pr1 0.24*** 0.80* 0.13 0.37 0.38 0.06 0.84** 0.16
pr2 0.45*** 0.82* 0.33* 0.13 0.56 0.44 −0.05 0.27
pr3 −0.19*** −0.81* 0.27 0.37* −0.36 0.54 0.84** 0.62**
r1 0.44*** −0.16 0.50 0.39* 0.54 0.06 0.84** 0.65**
r2 0.30*** 0.32* 0.62* 0.14 0.73* 0.46 −0.05 0.32**
r3 0.46*** −0.21 0.51 0.39* 0.59 0.56 0.84** 0.65**
Hypothesis 3 DEV ¼ b1 NBF þ b2 SER þ b3 NBF SER þ a
Multiple R 0.9*** 0.51** 0.85 0.82 0.57 0.82 0.54 0.5
R Squared 0.82 0.26 0.73 0.68 0.32 0.68 0.3 0.25
Adjusted 0.81 0.18 0.61 0.57 −0.08 0.43 0.14 0.18
R Squared
F(df1,df2) 102.68(3,7) 3.12(3,26) 6.28(3,7) 6.38(3,13) 0.8(3,5) 2.78(3,8) 1.96(3,4) 3.38(3,5)
p 0.0001 0.043 0.021 0.007 0.545 0.110 0.262 0.111
b1 7.01*** −2.19 1.70** 0.26 0.31 1.73 0.77 0.40
b2 0.89*** 0.41** 0.11 0.31** 0.03 −1.75 0.19 0.28
b3 −7.01* 1.92 −1.52** 0.12 0.28 0.96 −0.31 −0.09
pr1 0.77*** −0.1 0.82** 0.26 0.27 0.32 0.34 0.38
pr2 0.87*** 0.27** 0.63 0.32** 0.02 −0.31 0.18 0.26
pr3 −0.77*** 0.09 −0.80** 0.12 0.15 0.70 −0.13 −0.08
r1 0.44*** −0.16 0.46** 0.35 0.52 0.52 0.52 0.44
r2 0.73*** 0.41** −0.02 0.38** 0.46 0.56 0.18 0.35
r3 0.42*** −0.13 0.04 0.34 0.52 0.80 0.45 0.26
Hypothesis 4 DEV ¼ b1 NBF þ b2 GIT SER þ b3 NBF GIT SER þ a
Multiple R 0.7*** 0.54** 0.01 0.57** 0.76* 0.76** 0.83* 0.92**
R Squared 0.49 0.29 0.0001 0.32 0.58 0.58 0.69 0.85
Adjusted 0.48 0.2 0 0.26 0.44 0.51 0.54 0.8
R Squared
F(df1,df2) 65.09(3,71) 3.31(3,28) n.s. 5.19(3,13) 4.09(3,5) 8.42(3,8) 4.47(3,4) 16.73(3,5)
p 0.0001 0.034 n.s. 0.014 0.082 0.007 0.091 0.005
b1 0.15*** −0.55** 0.13 0.54** 0.32 0.1 0.79* 0.33
b2 0.66*** 0.46** −0.04 0.2 −0.25 0.34 0.1 0.15
b3 0.26*** −0.26 0.02 0.24 0.76* 0.73** 0.11 0.88**
pr1 0.20*** −0.09 0.13 0.57** 0.17 0.13 0.83* 0.62
pr2 0.70*** 0.03 −0.04 0.24 −0.05 0.24 0.18 0.19
pr3 0.23*** 0.05 0.02 0.24 0.76* 0.76** 0.2 0.92**
r1 0.39*** −0.29** 0.14 0.59* 0.75* 0.52** 0.87* 0.87**
r2 0.73*** 0.47** −0.05 0.37 0.75* 0.79** 0.24 0.88**
r3 0.69*** 0.12 0.02 0.53 0.76* 0.80** 0.33 0.97**
Symbols: standardized linear regression coefficients (b), measures of matching regression models (Multiple R, R Squared, Adjusted R Squared), analyses of
variance (Fisher’s test statistics F, observed probability p), partial linear correlation coefficients (pr), and Pearson’s linear correlation coefficients (r).
Statistically significant indexes are marked: 1) * for p < .1, 2) ** for p < .05, 3) *** for p < .01
Comment: Calculated bi value, the standardized regression coefficient, is reported from the estimated linear regression model, it is used as the effect size.
Standard error is obtained from test statistic to test the hypotheses: bi = 0. If an analysis reports the results of a simple linear regression Y = a0 + a1* X, the
effect size value b was calculated applying formula: b = (SD(X)/SD(Y))*a1, where SD(Y) is the standard deviation of response variable in the study and SD(X)
is the standard deviation of exposure measure used in the study. Mentioned formula enables to transform standardized effect into the effect related to
source data.
In our research, we were interested in demonstrating Variable a is a constant equal to 0 because our variables
that GIT (H2), SER (H3) and their interaction are standardized to the T-scale.
GIT*SER (H4) are moderators for correlation between The basic dependence is illustrated in Figure 3 by the
NBF and DEV. For this purpose, we used a multiple arrow b1 – the significance of the relation of the basic
linear regression method. In our study, moderators are variables NBF and DEV is an indicator of the existence
variables that affect the strength of the relation between of the basic dependence. A testimony that a third vari-
the predictor (NBF) and the criterion variable (DEV). able is a moderator of this dependence is the signifi-
Moderators specify when a relation will hold. The cance of arrow b3. However, it does not matter whether
model for testing moderation is shown in Figure 3. the relationship depicted by the arrow b2 is significant
This moderation can be presented each time as or not. The indicator that variable GIT has the status of
a regression model moderator of the correlation between NBF and DEV is
a significant interaction of variables NBF and
ðDEV ¼ b1 Predictor þ b2 Moderator þ b3 Moderator (GIT, or SER or their interaction GIT*SER).
Predictor Moderator þ aÞ Thus, we assumed in this equation: if the interaction
between the independent variable and a third variable (for
where b1, b2, b3 are regression coefficients, and a is example,b3 NBF GIT) is not statistically significant,
constant. Our criterion variable is DEV each time. then the third variable (in this example GIT) is not
Moderators are GIT, SER or their interaction GIT*SER. a moderator variable – it is just an independent variable.
Figure 3. Model for testing moderation.
Comparison of the mean DEV, NBF, GIT and SER in groups of

countries and total
DEV NBF GIT SER
70
60
50
T-Score
40
30
20
10
0
Advanced Emerging and Latin America Commonwealth Emerging and Middle East. Sub-Saharan Total
economies Developing and the of Independent Developing North Africa. Africa
Europe Caribbean States Asia and Pakistan
Figure 4. Comparison of the mean DEV, NBF, GIT and SER in individual groups of countries in relation to the total.
If b3 is statistically significant, then the third variable (e.g., First, we investigate GIT as a moderator for the
GIT) will be a moderator variable, and thus moderation is correlation between NBF and DEV. According to the
supported. The absolute values |bi| of the coefficients aforementioned idea of moderation, we assumed that
show which variables have more influence on the depen- there is a linear correlation between NBF and criterion
dent variable. Ratios with a higher absolute value are variable DEV with GIT as a moderator. We can predict
more important in the prediction (Andreasen, 1988; the average values of the dependent variable DEV on
Ferguson & Takane, 1989). the basis of concrete values of independent variables as
Table 2 contains the results of multiple linear regres- NBF, GIT and their interaction NBF*GIT.
sion analysis conducted at the country level and on This moderation can be presented as a regression
a global scale. It shows whether regression coefficients model:
b1, b2, and b3 are significant. Each of the standardized DEV ¼ b1 NBF þ b2 GIT þ b3 NBF GIT þ a
regression coefficients contain information on how much
a dependent variable increases or decreases, if the inde- where b1, b2, b3 are regression coefficients, and a is
pendent variable increases by the value equal to 1. The constant.
multiple correlation coefficient (R) indicates how strong Thus, a multiple linear regression was calculated to
the relationship is (Andreasen, 1988; Blalock, 1972; Shieh, predict in global DEV based on NBF, GIT, and
2010) between the group of independent variables (pre- NBF*GIT, and to investigate whether GIT can be trea-
dictors and moderators) and the dependent criterion ted as moderator. Table 2 (Hypothesis 2) shows that
variable DEV. The measure of regression model fit as an a significant regression equation was found (F(df1,
Adjusted R Squared (e.g., R2 = 0.38) indicates what part of df2) = 19.41, p < .01), with an adjusted R squared of
the variance of the dependent variable DEV can be 0.36. The multiple regression coefficient R = 0.6 indi-
explained or forecasted by the variables included in the cated medium strength of dependence between criter-
model. In our example, (100*R2 =), 38% of the change- ion variable DEV and the group of three independent
ability of development outcomes is explained by variables variables (NBF, GIT, and NBF*GIT). Thus, we can
in the model. Relatively high values of F Fisher’s empirical predict DEV as equal to:
statistics (Andreasen, 1988; Ferguson & Takane, 1989) DEV ¼ 1:70 NBF þ 0:58 GIT 1:35 NBF GIT
and the corresponding observed low probability level
p – close to zero – indicates that the model is significant This equation shows that DEV increased 1.70 standard
(Andreasen, 1988). The F test statistic indicates the ratio deviations for each increase of NBF by 1 standard
of the variance explained by the independent variables to deviation, and DEV increased 0.58 standard deviations
the variance of the random error of the model. The F test for each increase of GIT by 1 standard deviation, and
verifies the hypothesis if the regression model is signifi- DEV decreased 1.35 standard deviations for each
cant (Andreasen, 1988; Ferguson & Takane, 1989) increase of the interaction NBF*GIT by 1 standard
We also calculated partial correlation coefficients pr deviation. At the global level, therefore, NBF, GIT,
(Baba, Shibata, & Sibuya, 2004) that measure the degree of and interaction NBF*GIT were significant predictors.
association between two random variables (e.g., DEV and The significance of interaction NBF*GIT means that
NBF, or DEV and GIT), with the effect of a set of con- GIT is the observable moderator of correlation between
trolling random variables removed. In other words, the NBF and DEV. Thus, the hypothesis H2 is supported.
partial correlation coefficients indicate what is the pure
share of a given variable in the explanation and prediction Moderator: employment in the service industry
of a dependent criterion variable (DEV). We used them to Hypothesis H3: Employment in the service industry is
compare the strength of partial linear regression coeffi- a moderator that affects the correlation between new
cients between groups of countries. business formation and development outcomes.
We were also interested in Pearson’s linear regres- The next step of our study was to show that SER is
sion coefficients r that indicated linear associations a second moderator for correlation between NBF and
between considered variables. DEV. Again, we assumed that there is a linear correla-
tion between NBF and criterion variable DEV but this
time – with SER as a moderator. We checked whether
Moderator: open source participation the dependent variable DEV could be predicted on the
Hypothesis H2: Open source participation is a moderator basis of concrete values of independent variables as
that affects the correlation between new business forma- NBF, SER, and their interaction.
tion and development outcomes. This moderation can be presented as a regression model:
DEV ¼ b1 NBF þ b2 SER þ b3 NBF SER þ a As can be seen from Table 2 (Hypothesis 4),
a significant regression equation was constructed (F
where b1, b2, b3 are regression coefficients, and a is (df1, df2) = 65.09, p < 0,01), with an adjusted
constant. R squared of 0.49. The multiple regression coefficient
This time a multiple linear regression was com- R = 0.7 indicated strong dependence between criterion
puted to predict in global DEV based on NBF, SER, variable DEV and the group of three independent vari-
and NBF*SER, and to test if the SER can be ables (NBF, GIT*SER, and NBF*GIT*SER). Thus, we
a moderator. As can be seen from Table 2 can forecast DEV as equal to:
(Hypothesis 3), a significant regression equation was
constructed (F(df1, df2) = 102.68, p < .01), with an DEV ¼ 0:15NBF þ 0:66GIT þ 0:26NBFSER
adjusted R squared of 0.82. The multiple regression This equation shows that DEV increased 0.15 stan-
coefficient R = 0.9 indicated very strong association dard deviations for each NBF standard deviation, and
between criterion variable DEV and the group of three DEV increased 0.66 standard deviations for each
independent variables (NBF, SER, and NBF*SER). GIT*SER standard deviation, and DEV increased 0.26
Thus, we can forecast DEV as equal to: standard deviations for each interaction NBF*GIT*SER
standard deviation. So, in global NBF, GIT*SER, and
DEV ¼ 7:01 NBF þ 0:89 GIT 7:01 NBF SER
interaction NBF*GIT*SER were significant predictors.
This equation shows that DEV increased 7.01 stan- Thus, also GIT*SER can be a moderator for the corre-
dard deviations for each NBF standard deviation, and lation between NBF and DEV.
DEV increased 0.89 standard deviations for each GIT As can be seen from Table 2 (Hypothesis 4) in global,
standard deviation, and DEV decreased 7.01 standard two moderators together GIT*SER (thus their interac-
deviations for each interaction NBF*GIT standard tion) influenced more strongly DEV than GIT as
deviation. So, in global NBF, SER, and interaction a moderator. This difference manifests by values of multi-
NBF*SER were significant predictors. The significance ple R and partial correlation coefficients (pr) that were
of interaction NBF*SER means that SER is the obser- higher in the case of the model with GIT*SER (R = 0,70)
vable moderator of correlation between NBF and DEV. as the moderator than in the case of GIT (R = 0,38) as the
Thus, the hypothesis H3 is supported. moderator. In contrast, the model with SER as the mod-
erator has a higher level of multiple R (R = 0,90) than the
model with two moderators together GIT*SER (R = 0,70).
Interaction of moderators This may be due to the fact that SER is correlated with
Hypothesis H4: The two moderators, open source parti- GIT, and in addition, SER is correlated with DEV more
cipation and employment in the services industry, have strongly than GIT (Table 3). Therefore, the SER modera-
a stronger influence together than each individually. tor may contain information about GIT. Thus, hypothesis
This section investigates if the two moderators GIT H4 is only partially supported in global and in individual
and SER, together, thus their interaction GIT*SER have groups of countries.
a stronger influence, than one individually. Thus, we
assumed that the interaction GIT*SER is a moderator
for correlation between NBF and DEV. We continued Testable statement concerning groups of countries
our way of thinking that there is a linear correlation
Testable Statement TS: The correlations of hypotheses
between NBF and criterion variable DEV but this
H1-H4 vary by group of countries.
time – with interaction GIT*SER as a moderator. We
Pearson’s linear correlation coefficients and partial
checked whether the dependent variable DEV could be
correlation coefficients (Table 2) indicates differences
predicted on the basis of concrete values of indepen-
between groups of countries. We want to discuss the
dent variables as NBF, and interaction GIT*SER.
differences in correlations by groups of countries in the
We considered the interaction of two moderators
context of Hypothesis H1-H4. The first question we
GIT and SER in following equation (Table 2.
Hypothesis 4):
Table 3. Pearson’s linear correlation coefficients for DEV, NBF,
DEV ¼ b1NBF þ b2NBFGITSER þ a GIT, SER in global. Significant coefficients with p < .05 are
marked with *.
Thus, again a multiple linear regression was calcu- DEV NBF GIT SER
lated to predict in global DEV based on DEV 1.00 0.39* 0.43* 0.70*
NBF 0.39* 1.00 0.06 0.47*
NBF; GITSER; NBFGITSER. We wanted test GIT 0.43* 0.06 1.00 0.29*
also if the GIT*SER can be a moderator. SER 0.70* 0.47* 0.29* 1.00
need to address is what groups of countries we should mean we found in Emerging and Developing Europe
use in our analysis. and Commonwealth of Independent States. In these
We want to compare groups of countries not countries (AE and CIS) were also high enough level of
based on preexisting classifications but based on New Business Formation. It is worth mentioning that
their level of development outcomes. We consider less matured economies are characterized by clearly
groups with a similar level of development outcomes lower levels of employment in the service sector
to share a similar type of economy. The taxonomic (SER). As can be seen from Figure 5, similar tenden-
method of k-means enabled us to divide the coun- cies characterize employment in the service sector
tries into groups of different types of economies. We (SER) and opensource tools using (GIT). However,
maximized the distance between groups (Cailloux, a lot of less matured economies such as Sub-Saharan
Lamboray, & Nemery, 2007), which we verified African countries or Emerging and Developing Asia
using ANOVA variance analysis. We used variance try to balance this difference by new business forma-
analysis to determine if certain independent variables tion (NBF) and by applying opensource tools (GIT).
(factors as types of economy) influenced the level of Surprisingly strong correlations (Figure 5) exist for
the dependent variable DEV. We identified seven DEV and SER in the groups Sub-Saharan Africa
groups, which we also compared in terms of the (r = 0.91), Emerging and Developing Asia (r = 0.8)
degree of the aforementioned socioeconomic devel- Commonwealth of Independent States (r = 0.66), and
opment (Table 5). A standardized t-score value can Latin America and the Caribbean (r = 0.56).
indicate a low (less than 40), a medium (more than We can also observe significant differences related
40 and less or equal to 60), and a high degree of to moderation of correlation between DEV and NBF
socioeconomic development. Table 4 shows the ana- by GIT not only in global but also in most groups of
lysis of variance results that all groups of countries countries (Table 2), excluding group 2: Emerging and
are significantly different. Developing Europe and group 4: Commonwealth of
Table 5 shows that the highest development out- Independent States (see Table 2 Hypothesis 1). The
comes (DEV) above mean results are found in variable GIT is a significant independent variable in
Advanced Economies. Next, a little higher than global group 2 and in group 4. In group 4, all partial
Table 4. The results of k-means taxonomic method. Groups of countries based on their type of economy and analysis of variance
between groups.
Degree of
socioeconomic
Group Type of Economy development Symbol Examples of countries from the group that reported all: NBF, GIT and SER N
1 Advanced economies Very high AE Australia, Austria, Belgium, Canada, Cyprus, Czech Republic, Denmark, Finland, 36
France, Germany, Iceland, Ireland, Israel, Italy, Japan, Latvia, Lithuania, Luxembourg,
Malta, Netherlands, New Zealand, Norway, Slovakia, Slovenia, Spain, Sweden,
Switzerland, United Kingdom
2 Emerging and Mostly High EDE Bulgaria, Croatia, Hungary, Macedonia, Montenegro, Romania, Serbia, Turkey 11
Developing Europe
3 Latin America and the Mostly High LAC Argentina, Brazil, Chile, Colombia, Costa Rica, Dominican Republic, El Salvador, 25
Caribbean Guatemala, Jamaica, Mexico, Peru, Suriname, Uruguay
4 Commonwealth of Mostly Medium CIS Armenia, Azerbaijan, Kazakhstan, Russia, Ukraine 9
Independent States
5 Emerging and Mostly Medium EDA Bhutan, India, Indonesia, Malaysia, Nepal, Philippines, Sri Lanka, Thailand 17
Developing Asia
6 Middle East, North Rather low or MENAP Jordan, Pakistan, Qatar, Tunisia 18
Africa, and Pakistan Differentiated
7 Sub-Saharan Africa Very low SSA Ghana, Mauritius, Namibia, South Africa, Uganda 34
Analysis of variance – differences between clusters
Model df Model Variance MS Error df Variance of Error MS F Statistics p value <
DEV 6,00 1270,21 130,00 18,35 69,23 0,001
GIT 6,00 312,81 134,00 130,66 2,39 0,03
NBF 6,00 303,40 91,00 67,37 4,50 0,001
NBF*DEV 6,00 896,57 87,00 47,83 18,74 0,001
SER 6,00 828,02 80,00 45,45 18,22 0,001
Symbols Meaning
Model df degrees of freedom for the model (for clusters
Model Variance MS variance of the model of k-means
Error df degrees of freedom of the error
Variance of Error MS variance of the error
F Statistics Fisher-Snedecor F Statitics
P value Observed probability level
Table 5. Means and mean ranks based on source values (standardized T-scale in parentheses) for the dependent and independent
variables in relation to types of economy.
Type of Economy N DEV IDI HDI LE MSCH GNI II Scor. II Eff. HC GCI NBF GIT SER
AE 34 61.2 20.97 0.9 80.96 12.02 39892 53.17 0.74 79.88 4.81 6.73 2039.3 5.08
(61.41) (63.18) (62.76) (61.63) (61.77) (61.89) (63.15) (57.81) (61.6) (58.66) (57.25) (56.66)
EDE 11 51.52 60 0.8 76.12 10.54 17015 37.69 0.65 72.71 4.16 4.03 262.6 2.8
(53.31) (55.5) (56.38) (55.82) (57) (49.88) (50.59) (51.34) (54.71) (48.78) (49.12) (51.51)
LAC 23 47.64 89.91 0.72 74.07 8.38 12547 31.39 0.57 66.74 4.03 2.78 213.1 2.78
(48.93) (49.61) (51.55) (53.37) (50.03) (47.54) (45.48) (45.96) (48.97) (46.88) (48.91) (49.13)
CIS 8 48.46 61.33 0.75 71.62 11.44 12686 33.81 0.63 75.08 4.4 1.87 483.9 1.87
(52.46) (55.24) (53.2) (50.43) (59.91) (47.61) (47.44) (50.25) (56.99) (52.46) (50.1) (47.39)
EDA 15 43.67 109.85 0.67 71.36 7.01 13123 34.73 0.66 64.59 4 0.77 560.6 0.8
(47.47) (45.69) (48.33) (50.11) (45.61) (47.84) (48.18) (51.99) (46.9) (46.43) (50.44) (45.29)
MENAP 18 46.26 83.88 0.72 73.44 7.52 29481 29.52 0.59 59.15 4.36 1.02 25.9 1.07
(49.36) (50.8) (51.63) (52.61) (47.25) (56.42) (43.96) (47.3) (41.67) (51.92) (48.08) 45.77)
SSA 32 41.55 141.1 0.53 59.83 5.4 5163 25.2 0.52 55.68 3.8 2.09 18.4 2.11
(40.26) (39.54) (38.87) (36.27) (40.42) (43.66) (40.45) (42.5) (38.34) (43.48) (48.04) (47.81)
Total 141 50.9 81.39 0.72 72.21 8.67 20160 37.49 0.63 67.85 4.25 3.59 636.1 2.87
(50.46) (51.29) (51.56) (51.13) (50.95) (51.53) (50.42) (50.31) (50.04) (50.22) (50.8) (50.66)
Pearson's linear correlation coefficients between

moderators and development outcomes
GIT SER GIT*SER
1
0.8
0.6
0.4
0.2
0
-0.2
-0.4
All (Global) Advanced Emerging and Latin America Commonwealth Emerging and Middle East. Sub-Saharan
economies Developing and the of Independent Developing North Africa. Africa
Europe Caribbean States Asia and Pakistan
Figure 5. Pearson’s linear correlation coefficients between criterion variable DEV and interaction of NBF and moderators: GIT, SER
and interaction GIT*SER in different groups of countries.
correlations for NBF, GIT, and NBF*GIT are strong next in group 7 (Sub-Saharan Africa), and in group 5
enough, but the number of cases equal to 7 is too (Emerging and Developing Asia) and in group 3
small to conduct full regression analysis and general- (Latin America and the Caribbean).
ize conclusions. However, we mention these tenden- The moderation of SER was observed in group 2:
cies not to avoid possible interesting correlations. As Emerging and Developing Europe, in group 5
can be seen from Table 2 the moderator GIT has the (Emerging and Developing Asia). In the mentioned
strongest positive influence on the correlation groups, partial correlation and global correlation
between NBF and DEV in group 6 (Middle East, indicated positive dependency supporting DEV.
North Africa, and Pakistan), next in group 7 (Sub- We can also observe (Table 2) that the moderation
Saharan Africa), in group 5 (Emerging and of interaction GIT*SER on correlation between DEV
Developing Asia), and in group 3 (Latin America and NBF in the Commonwealth of Independent
and the Caribbean). The coefficients of regression of States (pr3 = 0.76), Emerging and Developing Asia
interaction between NBF*GIT are negative in the case (pr3 = 0.76), and in Sub-Saharan Africa (pr3 = 0.92)
of group 1: advanced economies and in global. Thus, is stronger than in the model with SER as moderator,
we can conclude that GIT as the moderator has the as well as in the model with GIT as the moderator.
greatest influence on socioeconomic development in In summary, our testable statement shows correlations
group 6 (Middle East, North Africa, and Pakistan), of hypotheses H1-H4 vary by certain groups of countries.
Discussion have strong implications for socioeconomic develop-

ment (Ardichvili et al., 2003). The importance of open
Implications for socioeconomic development
source participation and level of employment in service
The above findings have implications for socioeco- jobs in stimulating socioeconomic development efforts is
nomic development. The first and most important crucial. This research discovered that the interaction
implication is in the development outcomes. In our between the two factors in our model provides better
analysis, we found that several indicators related to predictability for the new business formation and devel-
development outcomes can be combined to a single opment outcomes.
development outcome construct. The new construct While this study has focused on OSS participation
for development outcome works well for most coun- in contributions to software development, the out-
tries and was shown to be a good measurement of comes from such OSS can change lives. One inspira-
overall development in a country. The development tional and well-documented case that motivates our
outcome construct simplified the analysis because it investigation was the SchoolNet Namibia that oper-
correlated to one construct and not many different ated in 2000–2009 to empower youth by bringing
outcome indicators. internet and technology to schools (Jacovkis et al.,
Secondly, implications for socioeconomic develop- 2010; Krakowski, 2006). SchoolNet Namibia was
ment vary between countries and their development a not-for-profit collaboration that used its limited
outcomes. We tested our model for different groups of resources to innovate on hardware, open source soft-
countries. An interesting finding is that countries with ware, and open educational resources to solve the
a well-developed economy are less reliant on new busi- immediate issues at Namibian schools. Employed
nesses for development outcomes. Alternatively, these open source software included the operating system
countries’ economies are so saturated that new business SUSE Linux, the office suite OpenOffice.org, and the
development forms a relatively smaller portion of their image editing software GIMP. Driven by principles of
development outcomes. In regions of the world where freedom, equality, diversity, inclusion, and self-
economies are less developed, the importance of new determination, SchoolNet Namibia trained volunteer
businesses for development outcomes is higher. learners and educators within schools to administer
A surprising finding was that emerging and developing the ICT. The newly learned ICT skills made previously
countries in Asia have fewer development outcomes unemployable volunteers employable. Additionally,
explained by new business formation. A possible expla- entrepreneurs used innovations from this open source
nation could be that geographical conditions and politi- project to establish internet cafés and generated
cal systems are less conducive for new entrepreneurs. income. Such an internet café would open a school
The scarcity of resources such as capital may be a factor library and its ICT after school hours to the public, sell
affecting the ability of microentrepreneurs to form new beverages, let clients surf the internet, and become
businesses in these countries. a place for the community to meet.
A key contribution of this research is that the level of The use of open source software for micro-loans
open source participation in a country impacts the illustrate our findings well. Micro-loans are an innova-
development outcomes and business formation. We tion for the development of small businesses operating
found that a country with higher open source partici- in low resource environments. The Grameen Bank
pation in GitHub enjoys higher development outcomes. received (The Nobel Peace Prize, 2006) for its efforts
Our theoretically derived hypothesis was supported on to help people in low-resource environments through
the global scale. In our analysis with countries of dif- micro-loans (The Nobel Peace Prize, 2006). Grameen
ferent economic situations, we found that the correla- open sourced its banking software Fineract to create
tion can be negative. In the countries of the Middle a community and support entrepreneurs that want to
East, North Africa, and Pakistan, the reverse correlation replicate the micro-loans business model (Mifos
was the strongest. The countries with Advanced Initiative, 2017c). A global open source community
Economies also had a negative correlation, although emerged with specialists and volunteers that collaborate
small, which might be explained by the availability of online and across multiple time-zones. The core tech-
funds to purchase licenses for proprietary software or nology (Fineract) is developed under the umbrella of
by the already widespread use of open source software. the Apache Software Foundation, which is a long-
Open source participation may affect development standing and respected open source community recog-
outcomes and business formation due to levels of entre- nized as a nonprofit organization.
preneurial alertness and opportunity identification. Ardichvili et al. (2003) suggested that in order for
Entrepreneurial alertness and opportunity identification innovation and growth to take place, entrepreneurs
need to identify business opportunities to create and occurs in smaller increments despite high levels of service
deliver value for stakeholders in prospective ventures. jobs. This is consistent with Fritsch and Mueller (2004)
They added that entrepreneurial opportunities are who found a correlation between new firm formation in
made, not found. An example of the creation of entre- the services sector and economic development. We found
preneurial opportunities through open source software a surprisingly small correlation in countries of the Middle
is the Apache Fineract open source community. The East, North Africa, and Pakistan. A possible explanation
Mifos Initiative facilitates a user community, supports could be that in these countries much outside investment,
an ecosystem, and fosters the volunteer network (Mifos especially from the US military, is used for improving the
Initiative, 2017b). The Mifos Initiative and Fineract infrastructure; and as such, the political system impacts
community software are used by 245 businesses to our data and makes these countries appear to be less
serve a total of 6,083,974 clients. This open source reliant on open source software than others.
community offers a platform for successful entrepre-
neurship and rise from poverty. Clients such as Marie-
Implications for businesses and governments
Claire Ayurwanda of Rwanda who survived HIV and
the death of two husbands was able to build a house for In this section, we offer business recommendations and
herself and her four children and started a second implications for research and practice that could con-
business in her village to employ others. Marie’s trium- tribute to the economic policy of companies and gov-
phant escape from poverty is what motivates this open ernments. While the adoption of technology can lead to
source community to keep the micro-loan finance insti- improvements in the lives of people, it appears that this
tutions and the borrowers they serve at the forefront alone cannot bring about improvements. Successful
(Mifos Initiative, 2017a). adoption of technology is often coupled with innova-
In countries with economic environments where tions, information literacy, and a host of social and
entrepreneurs and microenterprises cannot rely on economic factors (Qureshi, 2012). For the use of open
financial government support, the only available soft- source software to enable development efforts to be
ware is open source, and as such, the importance of successful, they will have to enable sets of behaviors
open source software is much larger in growing newly that are conducive to entrepreneurship and successful
founded businesses and bringing about development. opportunity identification.
The Mifos user community includes bank customers Businesses and governments can foster innovative
that are using the Fineract software of a micro-loan uses of open source software by microenterprises thus
financial institution. The community shares knowl- stimulating job creation in the services sectors and
edge, provides support, and influences the develop- growth. This innovativeness can enable increased com-
ment of the software. The Mifos Initiative certifies petitiveness, growth in enterprises, and the economy at
reliable partners that help entrepreneurs and financial large. Efforts by local or regional governments to
institutions adopt the Fineract technology. The Mifos ensure adequate ICT skill and training programs enable
Initiative states: “Our open source community-driven microenterprises to use open source software to bring
development and distribution promotes new business about growth. When microenterprises in a community
models that create and share value across our com- or region can see growth from open source software
munity rather than capture and contain it within one and better-trained people that they can hire, growth for
organization” (Mifos Initiative, 2017b). the communities and regions in which they reside can
The socioeconomic implications for supporting higher take place.
levels of employment in service jobs means that new The increased transparency afforded by open source
business formation and development outcomes will also software can reduce corruption at the national level
rise. We found a positive correlation between the level of and consequently at the local level in the interaction
employment in service jobs and development outcomes. with businesses and citizens because wrongdoers will
Interestingly, the correlation was positive for all groups of more likely be held accountable (Srivastava, Teo, &
countries. The countries with the highest correlation were Devaraj, 2016). While the use of any software may
in the Commonwealth of Independent States, Emerging not be directly connected to levels of corruption in
and Developing Asia, and Sub-Sahara Africa. A possible a country, it has been stated that the use of open source
explanation could be that in these countries the potential software in public-private partnerships is important
to grow service businesses is unexploited and provides because corruption is linked to lower levels of ICT
great potential for development to occur. In contrast, capacity in a country and can impede on the positive
countries with an Advanced Economy already have effects that ICT may have (Andoh-Baidoo, Osatuyi, &
sophisticated service industry jobs, and development Kunene, 2014). The protection of property rights is an
important element of development. Strict intellectual development outcomes rise with open source participa-
property laws can adversely affect access to medicines tion. In building upon Link and Qureshi (2017), this
and public health in low-resource nations (Owoeye, paper adds to what is known about the relationship
2016). Owoeye argues for a development-oriented between open source participation, the services sector,
approach for implementing intellectual property laws new business formation, and development outcomes. It
that will “enhance local pharmaceutical innovation, adds to the work of and Wennekers et al. (2005) by
easier access to essential drugs, and human develop- showing how open source participation in the service
ment” (p. 232). sector affects entrepreneurship through new business
Businesses and Government can work together to formation. This analysis is more realistic in that it
ensure that open standards do not put restrictions on measures development outcomes as an aggregate and
the use of the standard, e.g., through trade secrets or in a future year after the open source participation.
patents. Governments demand open standards to be The third contribution is to demonstrate that this
supported by software to ensure interoperability relationship is true on a global level but that some
between government agencies and the people. This groups of countries have an inverse correlation. Open
enables microentrepreneurs to offer their services at source participation has the strongest positive influ-
reduced costs to them and their customers. Open ence on the correlation between new business forma-
standards also reduce vendor lock-in and provide tion and development outcomes in countries that are
a level playing field for open source software and emerging and developing and almost negative in the
proprietary software, ensuring freedom of choice case of advanced economies. This means that open
and enabling innovation. Policymakers who find it source participation as a moderator has the greatest
challenging to lobby for open source software may influence on socioeconomic development in the
find it easier to lobby for open standards because it Middle East, North Africa, Sub-Saharan Africa,
does not exclude existing software vendors but only Emerging and Developing Asia and in Latin America
pushes them to openness. and the Caribbean. This offers opportunities for inves-
tigating growth in these areas.
The fourth contribution is the discovery that the
Conclusion and contributions level of employment in the services sector is a strong
The majority of studies in this area offer contributions moderator and displays a strong correlation with devel-
at the organizational level while regional and global opment outcomes. We find that this may be due to the
studies measure aggregate development (Qureshi, fact that the services sector variable (SER) is correlated
2015). While studies have shown that a linear correla- with open source participation (GIT). This finding is an
tion between new business formation and development important contribution to our understanding of open
outcomes may not be the best model (Wennekers et al., source participation and service sector growth.
2005), the contribution of this research is in that it
investigates the effects of an innovative use of technol-
Limitations and future work
ogy, open source software at the global level, while
offering specific insights into differences between coun- This work is limited by relying on publicly available
tries that are grouped based on their development out- data sources from international agencies and specifi-
comes. This paper makes several contributions to the cally GitHub for open source data. The data was not
development literature: collected for our purpose and as such might not be best
The first contribution is the conceptualization and suited for this analysis. Another limitation is the lack of
formalization of a development outcome construct available data for some countries which led to their
which contains economic, human, and social develop- exclusion from this study. The data analysis had several
ment variables. We demonstrate that this development surprises, which we think are related to geographical
outcome indicator is strongly correlated and can be and political circumstances including taxes or the pre-
combined into one convenient development outcome sence of special ethnic and religious groups in coun-
construct. We used this construct in this analysis and tries. This study did not explore these potentially
documented its development for other researchers to influential factors but encourages future work to inves-
use in their own research. tigate the qualitative foundation of our findings.
The second contribution is a confirmation that the While we were able to show a correlation between
level of open source participation has a positive corre- open source participation, we cannot claim to have
lation, independent of employment in the service shown causation. Future attempts to disprove our
industry. This means that new business formation and hypotheses may find a viable approach by investigating
how much entrepreneurial activity occurs through open Jolanta Kowal is Assistant Professor at the University of
source participation. It may be possible that open Wrocław, Institute of Psychology, Poland, Jungian analyst
source participation and new business formation thrive (IAAP), and Senior President of Polish Chapter of AIS. Her
research interests include methodology of socio-economic
in the same environments and by identifying the shared sciences, economic psychology, management, and multicultural
antecedents our hypothesis might be disproven, but in influences of psychoanalysis. She published in the journal of
the process, new ways to foster development might be Information Systems Management, Information Technology for
discovered. Whether entrepreneurs who used open Development, the Electronic Journal of Information Systems in
source software would have formed a new business if Developing Countries, the Journal of Global Information
Technology Management, and in proceedings of AMCIS,
there was no open source or whether open source has
HICSS, IEEE, and ICTM. She is an associate editor for the
become a critical factor for new business formation is journal Information Technology for Development, for the journal
another interesting question for future work. Similarly, Information Systems Management, among many others.
investigating the percentage of all new business forma-
Sajda Qureshi is Kayser Professor and Director of the
tions that relied on open source software is an intri- Information Technology for Development Cloud Computing
guing question. Future work may also consider other Lab at the College of Information Science and Technology at the
platforms where open source participation occurs and University of Nebraska at Omaha; Editor-in-Chief of the
expand the analysis to include additional forms of Journal of Information Technology for Development; and
participation and how they are related to entrepreneur- President of the AIS Special Interest Group for Global
Development (GlobDev). Since 10 years, she co-chairs the
ial activities.
Annual ICIS SIGGlobDev workshop and HICSS ITD mini-
track. She secured 1.1 million dollars in grants and contracts.
She has about 200 publications including in IEEE Transactions
Notes
in Professional Communication, Group Decision and
1. Information about approved open source licenses: Negotiation, Information Infrastructure and Policy, and
https://opensource.org/licenses. Communications of the ACM.
2. The count has probably increased by several million by
the time of publication. https://github.com/about.
3. SourceForge has been around longer but has lost popu- ORCID
larity to GitHub. https://sourceforge.net/about.
Georg J. P. Link http://orcid.org/0000-0001-6769-7867
Jolanta Kowal http://orcid.org/0000-0002-6241-9603
Sajda Qureshi http://orcid.org/0000-0002-7407-8148
Acknowledgments
The authors thank the special issue editors and anonymous
reviewers for their detailed and thoughtful comments on References
earlier multiple versions of this paper. In particular, the
authors are grateful to Narcyz Roztocki and Renata Acs, Z. J., Desai, S., & Hessels, J. (2008). Entrepreneurship,
Gabryelczyk for offering multiple rounds of detailed feedback economic development and institutions. Small Business
to help us strengthen the contributions of the research pre- Economics, 31(3), 219–234. Retrieved from http://www.
sented in this paper. jstor.org/stable/40650942
Andoh-Baidoo, F. K., Osatuyi, B., & Kunene, K. N. (2014). ICT
capacity as the investment and use of ICT: Exploring its ante-
cedents in Africa. Information Technology for Development, 20
Conflict of interest
(1), 44–59. doi:10.1080/02681102.2013.804399
The authors declare that they have no conflict of interest. Andreasen, A. R. (1988). Cheap but good marketing research.
Homewood, Illinois: Dow Jones-Irwin.
Ardichvili, A., Cardozo, R., & Ray, S. (2003). A theory of
Notes on contributors entrepreneurial opportunity identification and
development. Journal of Business Venturing, 18(1),
Georg J. P. Link earned his Ph.D. from the University of 105–123. doi:10.1016/S0883-9026(01)00068-4
Nebraska, USA. He contributed to open source for 13+ years Argyris, C., & Schön, D. A. (1978). Organizational learning:
and co-founded the Linux Foundation CHAOSS project A theory of action approach. Reading. MA, USA: Addision
(Community Health Analytics Open Source Software). Link Wesley.
holds a M.Sc. in Business Information Systems from the Avgerou, C. (2008). Information systems in developing countries:
Technische Universität Braunschweig, Germany, an MBA A critical research review. Journal of Information Technology,
from the University of Nebraska at Omaha, USA, and a B. 23(3), 133–146. doi:10.1057/palgrave.jit.2000136
A. in Business Administration from the WelfenAkademie, Baba, K., Shibata, R., & Sibuya, M. (2004). Partial correla-
Germany. Link published in the Communications of the tion and conditional correlation as measures of condi-
Association for Information Systems, the Journal of Universal tional independence. Australian & New Zealand Journal
Computer Science, and the Journal of Peer Production. Link of Statistics, 46(4), 657–664. doi:10.1111/j.1467-
received the AIS Doctoral Student Service Award 2017. 842X.2004.00360.x
Bagozzi, R., & Yi, Y. (2012). Specification, evaluation, and Donner, J. (2006). Internet use (and non-use) among urban
interpretation of structural equation models. Journal of the microenterprises in the developing world: An update from
Academy of Marketing Science, 40(1), 8–34. doi:10.1007/ India. In Conference of the Association of Internet
s11747-011-0278-x Researchers (AoIR).Brisbane, Australia.
Baron, R. M., & Kenny, D. A. (1986). The moderator– Duncombe, R., & Heeks, R. (2002). Enterprise across the
Mediator variable distinction in social psychological digital divide: Information systems and rural microenter-
research: Conceptual, strategic, and statistical considera- prise in Botswana. Journal of International Development,
tions. Journal of Personality and Social Psychology, 51(6), 14(1), 61–74. doi:10.1002/jid.869
1173–1182. doi:10.1037/0022-3514.51.6.1173 Feller, J., Finnegan, P., Fitzgerald, B., & Hayes, J. (2008).
Bharati, P., & Chaudhury, A. (2009). SMEs and competitive- From peer production to productization: A study of
ness: The role of information systems. International socially enabled business exchanges in open source service
Journal of E-Business Research, 5(1), i–ix. networks. Information Systems Research, 19(4), 475–493.
Bhatt, P., Ahmad, A. J., & Roomi, M. A. (2016). Social doi:10.1287/isre.1080.0207
innovation with open source software: User engagement Ferguson, G. A., & Takane, Y. (1989). Statistical analysis in
and development challenges in India. Technovation, 52–53, psychology and education (6th ed.). New York, NY:
28–39. doi:10.1016/j.technovation.2016.01.004 McGraw-Hill, Inc.
Blalock, H. M. (1972). Social statistics. New York, NY: Fitzgerald, B. (2006). The transformation of open source
McGraw-Hill. software. MIS Quarterly, 30(3), 587–598. doi:10.2307/
Bonaccorsi, A., Giannangeli, S., & Rossi, C. (2006). Entry stra- 25148740
tegies under competing standards: Hybrid business models in Fritsch, M., & Mueller, P. (2004). The effect of new firm
the open source software industry. Management Science, 52 formation on regional development over time (Freiberg
(7), 1085–1098. doi:10.1287/mnsc.1060.0547 Working Papers No. 2004,01). TU Bergakademie Freiberg,
Breitenfellner, A., & Hildebrandt, A. (2006). High employ- Faculty of Economics and Business Administration.
ment with low productivity? The service sector as Retrieved from https://ideas.repec.org/p/zbw/tufwps/
a determinant of economic development. Monetary Policy 200401.html
and Economy, Q1, 110–135. Garzarelli, G., Limam, Y. R., & Thomassen, B. (2008). Open
Brown, A. E., & Grant, G. G. (2010). Highlighting the duality source software and economic growth: A classical division
of the ICT and development research agenda. Information of labor perspective. Information Technology for
Technology for Development, 16(2), 96–111. doi:10.1080/ Development, 14(2), 116–135. doi:10.1002/itdj.20092
02681101003687793 Gerber, S. (2016, October 3). How startups can use open
Cailloux, O., Lamboray, C., & Nemery, P. (2007). source software to compete against the big guys.
A taxonomy of clustering procedures. In 66th Meeting of Retrievedfrom https://thenextweb.com/entrepreneur/2016/
the European Working Group on MCDA. Marrakech, 10/03/how-startups-can-use-open-source-software/
Morocco. Retrieved from https://hal.archives-ouvertes.fr/ Germonprez, M., Kendall, J. E., Kendall, K. E., Mathiassen, L.,
hal-00985860 Young, B., & Warner, B. (2016). A theory of responsive
Castelluccio, M. (Ed.). (2008). Enterprise open source design: A field study of corporate engagement with open
adoption. Strategic Finance, 90(5), 57–58. source communities. Information Systems Research, 28(1),
Chapman, R. L., Soosay, C., & Kandampully, J. (2002). 64–83. doi:10.1287/isre.2016.0662
Innovation in logistic services and the new business Grosh, B., & Somolekae, G. (1996). Mighty oaks from little
model: A conceptual framework. Managing Service acorns: Can microenterprise serve as the seedbed of
Quality: An International Journal, 12(6), 358–371. industrialization? World Development, 24(12), 1879–1890.
doi:10.1108/09604520210451849 doi:10.1016/S0305-750X(96)00082-4
Cosentino, V., Luis, J., & Cabot, J. (2016). Findings from Gruber, M., & Henkel, J. (2006). New ventures based on
GitHub: Methods, datasets and limitations. In Proceedings open innovation – An empirical analysis of start-up
of the 13th International Conference on Mining Software firms in embedded Linux. International Journal of
Repositories (pp. 137–141). New York, NY, USA: ACM. Technology Management, 33(4), 356–372. doi:10.1504/
doi:10.1145/2901739.2901776 IJTM.2006.009249
Dahlander, L., & Magnusson, M. G. (2005). Relationships ITU. (2017). Global ICT development index 2016. Retrieved
between open source software companies and commu- from http://www.itu.int/net4/ITU-D/idi/2016/
nities: Observations from Nordic firms. Research Policy, Jacovkis, D., Komen, J., & Tucker, K. (2010). The case of
34(4), 481–493. doi:10.1016/j.respol.2005.02.003 SchoolNet Namibia. Retrieved from http://wikieducator.
Dahlander, L., & Wallin, M. W. (2006). A man on the inside: org/The_Case_of_SchoolNet_Namibia
Unlocking communities as complementary assets. Research Jha, S. K., Pinsonneault, A., & Dubé, L. (2016). The evolution
Policy, 35(8), 1243–1259. doi:10.1016/j.respol.2006.09.011 of an ICT platform-enabled ecosystem for poverty allevia-
Davidov, E. (2008). A cross-country and cross-time compar- tion: The case of eKutir. MIS Quarterly, 40(2), 431–446.
ison of the human values measurements with the second Kelty, C. M. (2008). Two bits: The cultural significance of free
round of the European social survey. Survey Research software. Durham, NC, USA: Duke University Press.
Methods, 2(1), 33–46. doi:10.18148/srm/2008.v2i1.365 Kowal, J., & Jasińska-Biliczak, A. (2016). Socio−demographic
Di Tullio, D., & Staples, D. S. (2013). The governance and factors of innovative capacity of IT professionals in small
control of open source software projects. Journal of regional enterprises in transition economies. In Proceedings
Management Information Systems, 30(3), 49–80. of the European and Mediterranean Conference on
doi:10.2753/MIS0742-1222300303 Information Systems (EMCIS 2016). Krakow, Poland: Social
Science Research Network. Retrieved from https://papers. The Mozilla Foundation, & Open Tech Strategies. (2018).
ssrn.com/abstract=2898319 Open source archetypes: A framework for purposeful
Kowal, J., & Paliwoda-Pękosz, G. (2017). ICT for global open source (p. 39). Retrieved from https://blog.mozilla.
competitiveness and economic growth in emerging org/wp-content/uploads/2018/05/MZOTS_OS_
economies: Economic, cultural, and social innovations Archetypes_report_ext_scr.pdf
for human capital in transition economies. Information Nagle, F. (2018). Learning by contributing: Gaining compe-
Systems Management, 34(4), 304–307. doi:10.1080/ titive advantage through contribution to crowdsourced
10580530.2017.1366215 public goods. Organization Science, 29(4), 569–587.
Kowal, J., & Roztocki, N. (2013). Information and communi- doi:10.1287/orsc.2018.1202
cation technology management for global competitiveness Nanne, I., Moshabela, M., Huynh, U., & Diop, P. M.
and economic growth in emerging economies. The (2015). When development breeds contempt: Case
Electronic Journal of Information Systems in Developing examples of community-driven health systems
Countries, 57, 1–12. Retrieved from http://www.is.cityu. initiatives. Development in Practice, 25(8), 1120–1130.
edu.hk/staff/isrobert/ejisdc/57-0.pdf doi:10.1080/09614524.2015.1082534
Krakowski, P. (2006). ICT and free open source software in Neukrug, E. S., & Fawcett, R. C. (2014). Essentials of testing
developing countries. In J. Berleur, M. I. Nurminen, & and assessment: A practical guide for counselors, social
J. Impagliazzo (Eds.), Social informatics: An information workers, and psychologists (3rd ed.). Stamford, CT, USA:
society for all? In remembrance of Rob Kling (Vol. 223, pp. Cengage Learning.
319–330). Boston, MA: Springer US. doi:10.1007/978- The Nobel Peace Prize. (2006). Retrieved from https://www.
0-387-37876-3_26 nobelprize.org/nobel_prizes/peace/laureates/2006/
Larsen, R. J., & Marx, M. L. (1986). An introduction to OECD/Eurostat. (2005). Oslo manual: Guidelines for collect-
mathematical statistics and its applications (3rd ed.). ing and interpreting innovation data (3rd). Paris, France:
Upper Saddle River, NJ, USA: Prentice-Hall Inc. OECD Publishing. doi:10.1787/9789264013100-en
Leong, C., Pan, S. L., Newell, S., & Cui, L. (2016). The Owoeye, O. (2016). Intellectual property, health, regionalism
emergence of self-organizing e-commerce ecosystems in and development: A third world’s perspective. European
remote villages of China: A tale of digital empowerment Journal of Sustainable Development, 5(4), 225–232.
for rural development. MIS Quarterly, 40(2), 475–A8. Retrieved from https://ecsdev.org/ojs/index.php/ejsd/arti
Lindman, J., & Rajala, R. (2012). How open source has chan- cle/view/399
ged the software industry: Perspectives from open source Porter, M., & Kramer, M. R. (2011). Creating shared value.
entrepreneurs, Technology Innovation Management Review, Harvard Business Review, 89(1/2), 1–17.
5–11. (January 2012: Open Source Business). https://timre- Qiang, C. Z.-W., Clarke, G. R., & Halewood, N. (2006). The
view.ca/article/510 role of ICT in doing business. In Information and commu-
Link, G. J. P., & Qureshi, S. (2017). The role of open source in nications for development - Global trends and policies.
new business formation: Innovations for development. In Washington, DC: The World Bank. http://documents.
Proceedings of the Twenty-third Americas Conference on worldbank.org/curated/en/876661468154168686/Main-
Information Systems. Boston. Retrieved from http://aisel. report
aisnet.org/amcis2017/ICTs/Presentations/18/ Qureshi, S. (2010). Driving development through innovations
Link, G. J. P., & Qureshi, S. (2018). The role of open source in information technology and its applications.
communities in development: Policy implications for Information Technology for Development, 16(4), 241–243.
governments. In Hawaii International Conference on System doi:10.1080/02681102.2010.522040
Sciences 2018 (HICSS-51) Proceedings (pp. 2404–2413). Qureshi, S. (2012). Discovering development from infor-
Hawaii. Retrieved from http://hdl.handle.net/10125/50190 mation technologies: Does open access to technology
Malaquias, R. F., Malaquias, F., & Hwang, Y. (2017). The role improve the lives of people? Information Technology for
of information and communication technology for devel- Development, 18(3), 181–184. doi:10.1080/02681102.
opment in Brazil. Information Technology for Development, 2012.702858
23(1), 179–193. doi:10.1080/02681102.2016.1233854 Qureshi, S. (2015). Are we making a better world with infor-
McClelland, G. H., & Judd, C. M. (1993). Statistical difficul- mation and communication technology for development
ties of detecting interactions and moderator effects. (ICT4D) research? Findings from the field and theory
Psychological Bulletin, 114(2), 376–390. doi:10.1037/0033- building. Information Technology for Development, 21(4),
2909.114.2.376 511–522. doi:10.1080/02681102.2015.1080428
Mifos Initiative. (2017a,February 28). 3 Billion Maries. Qureshi, S., Kamal, M., & Wolcott, P. (2009). Information
Retrieved from http://mifos.org/impact/social-mission- technology interventions for growth and competitiveness
success/ in micro-enterprises. International Journal of E-Business
Mifos Initiative. (2017b, February 28). Global community. Research, 5(1), 117–140. doi:10.4018/978-1-61520-627-8.
Retrieved from http://mifos.org/mifos-initiative/global- ch015
community/ Riemenschneider, C. K., Harrison, D. A., & Mykytyn, P. P. (2003).
Mifos Initiative. (2017c,February 28). Our story. Retrieved Understanding IT adoption decisions in small business:
from http://mifos.org/about-us/our-story/ Integrating current theories. Information & Management, 40
Mitra, S. (2009, February 6). Open source means business. (4), 269–285. doi:10.1016/S0378-7206(02)00010-1
Retrieved from https://www.forbes.com/2009/02/04/open- Roztocki, N., & Weistroffer, H. R. (2016).
source-entrepreneur-technology-enterprise-tech_0206_mitra. Conceptualizing and researching the adoption of ICT
html and the impact on socioeconomic development.
Information Technology for Development, 22(4), von Hippel, E. (2002). Working paper 4366-02: Open source
541–549. doi:10.1080/02681102.2016.1196097 projects as horizontal innovation networks - by and for
Sahay, S. (2016). Are we building a better world with ICTs? users. MIT Sloan School of Management. Retrieved from
Empirically examining this question in the domain of public http://papers.ssrn.com/sol3/papers.cfm?abstract_id=
health in India. Information Technology for Development, 22 328900
(1), 168–176. doi:10.1080/02681102.2014.933522 von Hippel, E., & von Krogh, G. (2003). Open source soft-
Sen, A. K. (1999). Development as freedom. New York, NY: ware and the “private-collective” innovation model: Issues
Anchor Books. for organization science. Organization Science, 14(2),
Shieh, G. (2010). Sample size determination for confidence 209–223. doi:10.1287/orsc.14.2.209.14992
intervals of interaction effects in moderated multiple Walsham, G. (2017). ICT4D research: Reflections on history
regression with continuous predictor and moderator and future agenda. Information Technology for Development,
variables. Behavior Research Methods, 42(3), 824–835. 23(1), 1–24. doi:10.1080/02681102.2016.1246406
doi:10.3758/BRM.42.3.824 Wegner, M. (1987). Creating new jobs in the service sector.
Srivastava, S. C., Teo, T. S. H., & Devaraj, S. (2016). You can’t The Annals of the American Academy of Political and
bribe a computer: Dealing with the societal challenge of Social Science, 492, 136–150. Retrieved from http://www.
corruption through ICT. MIS Quarterly, 40(2), 511–A9. jstor.org/stable/1045284
Stec, M., Filip, P., Grzebyk, M., & Pierscieniak, A. (2014). Wennekers, S., van Stel, A., Thurik, R., & Reynolds, P. (2005).
Socio-economic development in the EU member states – Nascent entrepreneurship and the level of economic
Concept and classification. Engineering Economics, 25(5), development. Small Business Economics, 24(3), 293–309.
504–512. doi:10.5755/j01.ee.25.5.6413 Retrieved from http://www.jstor.org/stable/40229424
Stiglitz, J. E. (2002). Competition and competitiveness in a new Wheeler, D. A. (2015, July 18). Why open source software/
economy (MPRA Paper No. 44944) (pp. 11–22). Vienna, free software (OSS/FS, FOSS, or FLOSS)? Look at the
Austria: Austrian Ministry for Economic Affairs and Labour. numbers! Retrieved from https://www.dwheeler.com/oss_
Retrieved from http://mpra.ub.uni-muenchen.de/44944/ fs_why.html
Stiglitz, J. E., Sen, A., & Fitoussi, J.-P. (2010). Mismeasuring our Yakunina, R. P., & Bychkov, G. A. (2015). Correlation ana-
lives: Why GDP doesn’t add up. New York, NY: The New Press. lysis of the components of the Human Development Index
Swan, N. M. (1985). The service sector: Engine of growth? across countries. Procedia Economics and Finance, 24,
Canadian Public Policy/Analyse De Politiques, 11, 344. 766–771. doi:10.1016/S2212-5671(15)00692-9
doi:10.2307/3551061 Yunis, M., El-Kassar, A.-N., & Tarhini, A. (2017). Impact of
Thompson, B. (2004). Exploratory and confirmatory factor ICT-based innovations on organizational performance:
analysis: Understanding concepts and applications. The role of corporate entrepreneurship. Journal of
Washington, DC, USA: American Psychological Enterprise Information Management, 30(1), 122–141.
Association. doi:10.1037/10694-000 doi:10.1108/JEIM-01-2016-0040
Appendix 1. Variables Chosen for Comparisons and Sources of Data
Variable Description Source of Data

Dependent Variable and its component indicators (2015)
DEV Development Outcomes 2015 Constructed from below indicators as described in analysis of
Hypothesis H1
IDI ICT Development Index: development of ICT organizational and http://www.itu.int/net4/ITU-D/idi/2015/
technical infrastructure
HDI Human Development Index http://hdr.undp.org/en/composite/HDI
LE Life Expectancy http://hdr.undp.org/en/composite/HDI
MSCH Mean Years ofSchooling http://hdr.undp.org/en/composite/HDI
GNI Economic Growth: Gross national income (GNI) in USD https://data.worldbank.org/indicator/NY.GNP.MKTP.CD
II Score Global Innovation Index https://www.globalinnovationindex.org/analysis-indicator
II Efficiency Innovativeness Efficiency https://www.globalinnovationindex.org/analysis-indicator
HC Human Capital Index https://www.weforum.org/reports/the-global-human-capital-
report-2017
GCI Global Talent Competitiveness Index https://www.insead.edu/news/2017-global-talent-
competitiveness-index-davos
Independent Variables and moderators (2011–2013)
NBF New Business Formation New business density (new registrations per 1,000 people ages
15–64): https://data.worldbank.org/indicator/IC.BUS.NDNS.ZS
GIT Moderator 1: Open Source Participation on GitHub GH Torrent, available from Google Big Query. Link & Qureshi
(2017) describe the procedure in detail.
SER Moderator 2: Employment in Services http://databank.worldbank.org/data/download/site-content
/WDI_CETS.xls
GC Groups of Countries Own classification as described in analysis to hypothesis H6.
Reference:
Link, G. J. P., & Qureshi, S. (2017). The role of open source in new business formation: Innovations for development. In
Proceedings of the Twenty-third Americas Conference on Information Systems. Boston. Retrieved from http://aisel.aisnet.org/
amcis2017/ICTs/Presentations/18/

Link&Kowal&Qureshi 2019

Uploaded by

Copyright:

Available Formats

Link&Kowal&Qureshi 2019

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Link&Kowal&Qureshi 2019

Uploaded by

Copyright:

Available Formats

Information Systems Management

ISSN: 1058-0530 (Print) 1934-8703 (Online) Journal homepage: https://www.tandfonline.com/loi/uism20

Open Source in Development: Enabling Business

Georg J. P. Link, Jolanta Kowal & Sajda Qureshi

To link to this article: https://doi.org/10.1080/10580530.2020.1696548

Published online: 27 Nov 2019.

Submit your article to this journal

View related articles

View Crossmark data

Full Terms & Conditions of access and use can be found at

Open Source in Development: Enabling Business and Services

Introduction and cloud services (Lindman & Rajala, 2012). Open

Table 2. Verified models of moderation in relation to research hypotheses.

Figure 3. Model for testing moderation.

Comparison of the mean DEV, NBF, GIT and SER in groups of

Pearson's linear correlation coefficients between

Discussion have strong implications for socioeconomic develop-

Appendix 1. Variables Chosen for Comparisons and Sources of Data

Variable Description Source of Data

You might also like