Topics and Treatments in Global Software Engineering Research

A Systematic Snapshot

Bilal Raza, Stephen G. MacDonell and Tony Clear

SERL, School of Computing & Mathematical Sciences, Auckland University of Technology,

Private Bag 92006, Auckland 1142, New Zealand

Keywords: Global Software Engineering (GSE), Distributed Software Development, Systematic Mapping.

Abstract: This study presents an analysis of the most recent literature addressing global software engineering (GSE).

The primary purpose is to understand what issues are being addressed and how research is being carried out

in GSE – and comparatively, what work is not being conducted. We examine the current state of GSE

research using a new Systematic Snapshot Mapping (SSM) technique. We analysed 275 papers published

between January 2011 and June 2012 in peer-reviewed conferences, journals and workshops. Our results

provide a coarse-grained overview of the very recent literature addressing GSE, by classifying studies into

predefined categories. We also follow and extend several prior classifications to support our synthesis of the

data. Our results reveal that, currently, GSE studies are focused on Management and Infrastructure related

factors rather than Human or Distance related factors, using principally evaluative research approaches.

Most of the studies are conducted at the organizational level, mainly using methods such as interviews,

surveys, field studies and case studies. We use inter-country network analysis to confirm that the USA and

India are major players in GSE, with USA-India collaborations being the most frequently studied, followed

by USA-China. Specific groups of countries have dominated the reported GSE project locations (and the

locations of research authors). In contrast, regions including Central Asia, South Asia (except India), Africa

and South East Asia have not been covered in these studies. While a considerable number of GSE-related

studies have been published they are currently quite narrowly focused on exploratory research and

explanatory theories. The critical research paradigm has been untouched, perhaps due to a lack of criteria

and principles for carrying out such research in GSE. An absence of formulative research, experimentation

and simulation, and a comparative focus on evaluative approaches, all suggest that existing tools, methods

and approaches from related fields are being tested in the GSE context. However, these solutions may not

scale to cover GSE-related issues or may overlook factors/facets specific to GSE.

1 INTRODUCTION

Global software engineering (GSE) is a growing

field as is clearly evident in the diversity of locations

involved and the rapidly increasing number of

published studies into GSE-related issues. As the

number of such studies increases it becomes

important to periodically summarize the work and

provide overviews of the results (Petersen et al.,

2008), as a means of reflection on what work is

being done and what gaps might exist. Various fields

have specific methodologies to carry out such

secondary studies (Petersen et al., 2008). Often in

these studies, at least in the software engineering

(SE) domain, the evidence pertaining to a specific

topic or research question is investigated over a

period of five to ten years. In this paper, we utilize a

different approach – we investigate the breadth of

topics covered over a short timeframe, an approach

we refer to as a systematic snapshot. This establishes

a baseline state that could be extended in a backward

or forward direction to analyse changes over time.

The systematic mapping (SM) method has been

widely used in medical research (Petersen et al.,

2008) and was first adopted in software engineering

research by Bailey et al. (2007). SM aims to provide

high-level analysis of relevant research literature by

classifying the work according to a series of defined

categories and visualizing the status of a particular

field of research (Mohd Fauzi et al., 2010, Petersen

et al., 2008). This technique has been used recently

in the GSE field (Steinmacher et al., 2012, Jalali and

Wohlin, 2010, da Silva et al., 2011, Mohd Fauzi et

Raza B., G. MacDonell S. and Clear T..

Topics and Treatments in Global Software Engineering Research - A Systematic Snapshot.

DOI: 10.5220/0004444700850096

In Proceedings of the 8th International Conference on Evaluation of Novel Approaches to Software Engineering (ENASE-2013), pages 85-96

ISBN: 978-989-8565-62-4

 2013 SCITEPRESS (Science and Technology Publications, Lda.)

al., 2010, Portillo-Rodríguez et al., 2012). In these

studies specific aspects of GSE research were

categorized (using guidelines presented in

(Kitchenham and Charters, 2007, Petersen et al.,

2008)). These studies considered between 24 and 91

papers (24, 91, 77, 70 and 66 studies, respectively)

up to the year 2010 in their final analyses. The

aspects of GSE analysed in these studies were

software configuration management, awareness

support, agile practices, project management, and

tools in GSE. All five studies therefore classified the

GSE literature from a relatively narrow perspective

but covering a wide temporal range. They were

published in well-known journals and conferences

and provide valuable contributions to the body of

GSE literature. In our study, we instead use a new

systematic mapping process called Systematic

Snapshot Mapping (SSM), briefly described in

section 3, to classify the very current global software

engineering literature.

The next section provides a brief background to

related studies, and section III describes our method.

In the subsequent section IV our results are

presented followed by a discussion of validity

threats in section V. In section VI we conclude this

paper and section VII conveys future work.

2 BACKGROUND AND RELATED

MAPPING STUDIES

Interest in software development carried out by

globally distributed, culturally and/or temporally

diverse teams arose with the advent of outsourcing

in the last two decades and it continues to increase

(Šmite et al., 2010). Its importance has led to the

specific area of research and practice called global

software engineering (GSE) (Šmite et al., 2010). In a

recent review Šmite et al. classified the empirical

GSE research, considering studies published

between 2000 and 2008, and presented the results in

two papers (Šmite et al., 2008, Šmite et al., 2010).

They concluded that GSE was (still) an immature

field with limited empirical studies. They further

concluded that the majority of studies focused on

different aspects of GSE management rather than the

in depth analysis of GSE solutions.

Jalali and Wohlin (2010) reported a SM study

based on their analysis of 77 studies published

between 1999 and 2009. They focused their work on

the application of agile practices in GSE and

explored under which circumstances these practices

have been used successfully in that context. The

results reveal that in most cases agile practices were

modified based upon the context and requirements.

The authors also expressed the need for integrating

experiences and practices to assist practitioners. da

Silva et al. (2011) presented an evidence-based

project management model for distributed software

development based on the synthesis of 70 papers

published between 1997 and 2009. They aimed to

provide feedback to help practitioners and

researchers understand challenges and implement

effective solutions to improve project management

in distributed settings. Fauzi et al. (2010) presented

the results of a SM study of software configuration

management (SCM) in GSE. They found that a lack

of group awareness and coordination exacerbates the

issues of SCM and no process had been proposed to

address this. Their review considered 24 papers

published between 1999 and 2010. Rodriguez et al.

(2012) conducted a SM study, analysing 66 papers

published between 2000 and 2010. They compiled a

list of 132 tools used in global software projects and

classified them to help practitioners and researchers

make use of the available tool support. It was found

that the majority of these tools had been developed

at research centres and just 19% were reported to

have been tested outside the context in which they

were developed. Another SM study was reported by

Steinmacher et al. (2012). In this paper they

reviewed 91 studies regarding awareness support in

distributed software development (DSD). They

found that coordination is the most supported

dimension of the 3C model whereas communication

and cooperation are less frequently explored. All of

the above mentioned SM studies provide valuable

contributions to the body of GSE literature and

include content intended to support practitioners.

Each addresses a specific aspect of GSE and

considers around a decade of research in the field. In

our study we covered a shorter time period using a

different approach, described in the next section.

3 METHOD AND CONDUCT

The results presented in this paper correspond to our

classification of the current literature on GSE. We

used a new method for carrying out SM studies

called Systematic Snapshot Mapping (SSM). In

order to classify the current literature, we chose the

time period between January 2011 and June 2012.

This study followed guidelines presented by Petersen

et al. (2008) for carrying out systematic mapping

studies. However, instead of narrowing down the

topic and considering a large temporal period, we

ENASE2013-8thInternationalConferenceonEvaluationofNovelSoftwareApproachestoSoftwareEngineering

limited the time span and considered the full breadth

of topics covered. This study was inspired by several

prior classifications of SE and GSE literature

including that of Glass et al. (2002), but instead of

following a random sampling technique to select

papers (as in (Glass et al., 2002)) we used a

systematic process. We followed the general

guidelines of (Petersen et al., 2008) and employed a

defined protocol for choosing search strings and

executing them against relevant databases to cover

the breadth of GSE-related studies. Thus, instead of

limiting the topic itself (as per the SMs cited above)

we limited the scope by using a small temporal

range, giving us an up-to-date snapshot overview of

the literature. We defined our categories at the outset

of our analysis and chose various dimensions to

present the results, mainly leveraging the prior

classification work of Richardson et al. (2012) and

Glass et al. (2002). We present our results in the

form of tables, bar graphs, bubble plots and network

analysis graphs to provide visual representations of

the data. We believe such a snapshot approach is

especially useful in cases where a field is moving

rapidly and where there is consequently rapid

growth in the research literature. This new approach

for carrying out SM also provides an opportunity to

effectively build upon different researchers’ work by

using different temporal ranges. Since traditional

approaches such as systematic literature reviews and

systematic mappings use narrowly defined topics it

is difficult to analyse how overall trends evolve over

a period of time. This study provides a baseline

against which analyses using other temporal ranges

could be compared.

3.1 Research Questions

In order to present a current snapshot of the GSE

research literature, the following research questions

were established for this study:

RQ1. What are the factors, levels and locations

investigated in the current GSE literature?

RQ2. How is the current research being carried

out in GSE in regard to methods and approaches?

3.2 Search Strategy

Our search strategy was designed to keep the topic

general while addressing a short time period to

provide an up-to-date overview of the research

literature. Initial search keywords were selected

from known GSE systematic literature reviews and

mapping studies. These keywords were updated

based upon various dry runs carried out on the

Scopus database to ensure their effectiveness. In the

initial run, a target was set to ensure at least those

studies from which the keywords were taken were

retrieved. In the second run, a random set of ten

studies was selected from the Proceedings of the

2009-2011 ICGSE conferences, and the search

strings were further refined to ensure that these

sample studies were also retrieved. A similar method

was used by Jalali and Wohlin (2010) to justify and

improve the utility of their selected key terms.

Table 1 shows the final list of keywords used to

cover as many variations of the same term as

possible. As this area of research is still maturing,

we intentionally adopted many keywords having low

precision but high recall (Dieste and Padua, 2007).

Table 1: List of keywords used as search strings.

3.3 Data Sources

We searched across multiple data sources to retrieve

as many potentially relevant studies as possible.

Initial preference was given to the use of the

electronic database Scopus as it provides

comprehensive coverage of relevant GSE journals. It

is especially recommended for the software

engineering and computer science fields as it covers

many of the well-known publishers in these

disciplines. Simultaneously, IEEE Xplore, the ACM

Digital Library, SpringerLink and ScienceDirect

were also searched to complement the Scopus

results. Each database has its limitations in terms of

the number of keywords accepted at a specific

instance; therefore, we had to break the search

phrases to suit the particular database. These

subsequent searches, which tended to find limited

additional studies to those found via Scopus, added

to our confidence that relevant studies had not been

missed in our search.

3.4 Data Retrieval

Data was retrieved in multiple steps. In the first step,

citations of retrieved studies were downloaded and

stored as separate EndNote files based upon their

abstracts and titles. After this process, all the

TopicsandTreatmentsinGlobalSoftwareEngineeringResearch-ASystematicSnapshot

EndNote files were combined and duplicate papers

were removed. Once all the duplicates were

removed the studies were then considered for the

inclusion process. The search and retrieval process

was conducted in July 2012 and the date range was

limited to January 2011 to June 2012. The search

was carried out on metadata (title, abstract,

keywords) and only peer-reviewed literature

published in English was considered.

3.5 Inclusion Process

The steps taken in the inclusion process to select

studies are shown in Figure 1. After searching each

database 2020 studies were retrieved. The decision

for further analysis of studies was based upon the

first author’s reading of the papers’ titles or abstracts

(resulting in 1125 studies). After this step, duplicates

were removed using the ‘Find Duplicates’ feature of

EndNote – this led to the removal of more than half

of the studies under consideration. After removing

the duplicates full text versions of each study were

sought. For 12% of the papers (53 of the 437

remaining) the full text was not available to us,

primarily because the papers were not published in

well-known journals or conference proceedings.

These studies were not considered for further

analysis. The full text of the remaining 384 papers

was then reviewed by the first author and a final set

of 275 studies was selected for inclusion in the SM

analysis. In this stage, studies in the form of short

papers, extended abstracts and position papers (only

describing future work) were excluded. A number of

studies, not related to the software engineering

domain, had slipped through to this stage and upon

cursory review of the full text were also excluded. .

3.6 Data Extraction and Synthesis

We followed the general guidelines provided in

(Petersen et al., 2008) to build a classification

scheme. The included studies were categorized

according to various dimensions: research approach,

research method, GSE factors, level of analysis and

GSE locations. In order to reduce the threats to

validity, regular meetings of the three authors were

held to discuss issues and address misconceptions.

In order to reduce bias effects the three researchers

also conducted a sample classification together. At a

later point a further sample of studies which were

initially classified by the first author were verified

by the senior researchers, discussions were held

again and issues were addressed. It was established

Figure 1: Inclusion process.

that the authors were in general agreement regarding

the classification, based upon the sample results.

The classification scheme utilized by Glass et

al. (2002) was used to characterize the research

approach for our set of studies. This scheme (Glass

et al., 2002) was based on an earlier categorization

by Morrison and George (1995) in which the

following four approaches were employed:

Formulative, Evaluative, Descriptive and

Developmental. The first three categories were

further divided to provide a rich set of candidate

research approaches (Glass et al., 2002). The

descriptive and formulative categories characterize

non-empirical studies (Glass et al., 2002). The

descriptive category has three subcategories.

Subcategory Descriptive-System was used to capture

papers focused on describing a system, Descriptive-

Other was used to categorize papers that included an

opinion piece, and the third descriptive category

used was Review of literature (Glass et al., 2002).

The formulative approaches were classified into six

subcategories to cover the major entities being

formulated: Framework, Guidelines/standards,

Model, Process/method, Classification/taxonomy

and Concept (Glass et al., 2002). The Evaluative

category in (Glass et al., 2002) drew on three main

research epistemologies identified by Orlikowski

and Baroudi in (1991): Positivist as Evaluative-

Deductive, Interpretive as Evaluative-Interpretive,

and Critical as Evaluative-Critical. Evaluative-Other

(Glass et al., 2002) was added in this list to include

studies that have an evaluative component but did

Final Search

Read abstracts/titles

Analysis

Removing duplicates

If Full text available

Text review

275

2020

1125

437

384

275

Peer reviewed, with in title, abstract, keywords,

From Jan ’11 to June ‘12

ENASE2013-8thInternationalConferenceonEvaluationofNovelSoftwareApproachestoSoftwareEngineering

not use any of the above approaches e.g. opinion

surveys. So in total, we used these 13 subcategories

formulated in (Glass et al., 2002) to classify the

research approaches used in our set of GSE studies.

It was found that epistemologies were rarely

mentioned in the Abstract or Introduction of the

studies and various other sections had to be

traversed to enable this particular classification.

Glass et al. (2002) also encountered this same issue.

We also mainly considered (Glass et al., 2002)

for the list of methodologies used in software

engineering research. However, to better reflect the

GSE perspective we also considered the

methodologies considered in (Šmite et al., 2008,

Šmite et al., 2010). We added computer mediated

communication (CMC) analysis to cover studies that

investigate artefacts such as chat-histories and

emails. Although grouped together in prior studies,

Observations and Interviews were considered

separately, as many studies use them to complement

other methods. Interviews are widely used as a sub-

method in Case Studies and Observations are used in

Ethnographies. However, we observed that these

methods are being used in their own right and we

therefore classified them separately. We included the

method Data Analysis to signify studies that utilized

data from Repositories, Incident Management

Systems and Archives of previous projects. We used

Proof of Concept for non-empirical studies in which

entities were formulated but were only described by

examples rather than any formal validation. Some

other generic data along with the above

categorizations were captured in a spread sheet:

 Title of the paper

 Author (s) and their geographic location

 Summary of the conclusion

 The source (journal, conference or workshop)

 Process activities/artefacts, practices (if

discussed)

 Type of contribution

 Geographical locations of mentioned projects.

4 FINDINGS

This section presents the results obtained based on

the data extracted from our final set of 275 studies.

4.1 Findings for Factors

Richardson et al. (2012) identified 25 GSE factors in

an empirical study and grouped them in the four

broad categories of Distance, Infrastructure,

Management and Human Factors. We used these

categories to also characterize our identified studies.

We added Learning/Training/Teaching, Competition

and Performance to the Management category and

Relationship to the Human Factors category. We

also updated the latter category with

Coordination/collaboration. Table 2 presents the

results of this classification. The results clearly show

that current GSE studies are heavily focused on

Management and Infrastructure related factors

compared to Human and Distance related factors.

Šmite et al. in (2010) presented a systematic review

of empirical GSE studies and also found that most of

the studies were focused on management related

issues. Comparing these results with the SWEBOK

(Alain et al., 2001) knowledge areas (KAs), it was

found that the standard lacks specific considerations

for GSE. As a corollary, it was also found that KAs

related to design, construction, testing and

maintenance are not widely addressed in the recent

GSE literature.

Table 2: Findings for GSE factors.

GSE Factors Percentage

Distance 17.5%

Communication 8.9%

Language 1.1%

Culture 5.5%

Temporal issues 1.8%

Human Factors 14.7%

Fear 0.5%

Motivation 2.3%

Trust 2.7%

Cooperation 1.8%

Coordination/collaboration 5.9%

Relationship 1.4%

Management 44.5%

True Cost 1.8%

Project Management 8.9%

Risk Management 2.3%

Defined roles and responsibilities 1.6%

Team Selection 0.9%

Effective Partitioning 4.6%

Skills Management 0.4%

Knowledge transfer/knowledge 6.7%

Visibility 3.4%

Reporting Requirement 0.0%

Information Management 1.1%

Teamness 5.5%

Learning/Training/teaching 4.6%

Competition 0.6%

Performance 1.6%

Infrastructure 23.3%

Process Management 7.1%

Tools 8.5%

Technical Support 0.4%

Communication tools 7.1%

TopicsandTreatmentsinGlobalSoftwareEngineeringResearch-ASystematicSnapshot

4.2 Findings for Research Approach

GSE presents a complex context that demands a

more extensive repertoire of research methods and

approaches than those currently prevailing (Clear

and MacDonell, 2011). Table 3 presents the findings

of the classification of research approaches used in

current GSE-related studies. In terms of the three

main categories, the dominant research approach is

Evaluative, followed by Descriptive and then

Formulative. This is in sharp contrast to the results

reported in 2002 by Glass et al. in which the order

was Formulative, Descriptive and Evaluative. One

of the main reasons for the present dominance of

Evaluative research is the inclusion of new empirical

methods such as CMC analysis, Interviews, Data

Analysis and Observations. These results appear to

be in contrast with the results of Šmite et al.’s

systematic review (2010) of GSE-related studies

published between 2000 and 2008. They concluded

that GSE-related studies are relatively small in

number and immature and most of them focused on

problem-oriented reports. Our current results show,

however, that GSE publications have grown in

quantity and quality and more studies have used

evaluative approaches. Of note is these evaluative

approaches are mostly confined to previously

formulated work. We interpret this to mean that

existing methods, tools and so on from related fields,

such as collocated software engineering (CSE), are

being evaluated in the context of GSE.

Given that GSE is fundamentally different from

CSE (Richardson et al., 2012), it seems likely that

solutions formulated for CSE will need to be

updated or enhanced for GSE. Entirely new

solutions may also need to be identified and assessed

in the GSE context. Similarly, there is clear potential

for critical research in this context particularly in

light of the power structures that can exist between

GSE ‘partners’, and the associated issues of trust,

fear, cooperation and the like (as shown in Table 2).

Criteria or principles for carrying out critical

research are lacking generally in information

systems (IS) (Myers and Klein, 2011). Considering

its importance, Myers and Klein in (2011) proposed

a set of principles for conducting critical research –

these principles could be considered in future

investigations of human factors in GSE.

4.3 Findings for Research Methods

Figure 2 depicts the research methods used. The

most dominant methods are Interview, Survey, Field

Study and Case Study, indicating that most of the

studies employed qualitative methods. These results

are also in stark contrast to (Glass et al., 2002) in

Table 3: Findings for research approach.

Research Approach Percentage

Descriptive 25.4%

Descriptive-system (DS) 7.8%

Review of literature (DR) 9.7%

Descriptive-other (DO) 7.8%

Evaluative 56.2%

Evaluative-deductive (ED) 17.1%

Evaluative-interpretive (EI) 26.4%

Evaluative-critical (EC) 0.0%

Evaluative-other (EO) 12.7%

Formulative 18.3%

Formulative-framework (FF) 5.1%

Formulative-guidelines/standards/approach 1.9%

Formulative-model (FM) 5.6%

Formulative-process, method, algorithm 3.1%

Formulative-classification/taxonomy (FT) 0.7%

Formulative-concept (FC) 1.7%

which SE researchers used very few case or field

studies. For studies in which multiple methods were

used we assigned more than one research approach

and method. Research methods in GSE are currently

skewed towards exploratory research focusing on

theories relating to ‘Explanation’ as described by

Gregor (2006). These theories aim to provide

explanation about what, how and why things happen

and to promote greater understanding of phenomena.

Thus, although GSE research has grown in terms of

the number of studies being conducted, these studies

are exploratory and/or explanatory in nature. It will

be interesting to compare these results with future

studies to determine whether work moves towards

more predictive studies as the field matures.

Figure 2: Findings for research methods.

ENASE2013-8thInternationalConferenceonEvaluationofNovelSoftwareApproachestoSoftwareEngineering

4.4 Findings for Level of Analysis

Figure 3 shows the level of analysis considered

currently by GSE researchers. The dominant level of

analysis was found to be Organizational followed by

Inter-Organizational - combined together they are

used in more than half the studies reviewed. Fewer

studies addressed group, individual and societal

levels, a finding that coincides with the results of

Glass et al. (2002) in respect of SE studies.

Figure 3: Findings for level of analysis.

4.5 Findings for Distribution of Studies

Table 4 presents the distribution of studies across

various conferences, journals and workshops with

frequency greater than one. (This limit was imposed

due to space considerations and for ease of

interpretation.) The majority of the selected studies

were published in conference proceedings and drew

on an industrial context.

4.6 Bubble Plot Analysis

The use of visual techniques, such as bubble plots,

has been recommended by (Petersen et al., 2008) and

such techniques have been used to convey the results

of mapping and classification studies (Šmite et al.,

2008, Jalali and Wohlin, 2010). Figure 4 presents the

results of this study in the form of a bubble plot. We

chose to represent three classifications within it:

Research approach is on the right X-axis, GSE-

factors, grouped in their four major categories, are on

the Y-axis, and level of analysis is on the left X-axis.

The results clearly show that most of the current

studies are focused on using evaluative approaches

around management and infrastructure factors and

analysed at the organizational levels. Studies based

Table 4: Distribution of studies across Journals,

Conferences and Workshops.

Journals

CSCW 8

IST Journal 8 PROFES 6

JSEP 7 CHI 5

J SOFTW MAINT EV 7 XP 4

IET Software 6 ICIC 3

J of E MARKETS 4 PICMET 3

IEEE Software 4 ISEC 3

J COMM and COM SC 3 ICSSP 3

ISJ 3 MySEC 2

IJoPM 2 EUROMICRO 2

JSW 2 ICIS 2

POM Journal 2 CollaborateCom 2

IS 2 CTS 2

IEEE TEM 2 PACIS 2

LNBIP 2

Workshops

J Grp Dec Negot 2 CTGDSD 13

Conferences

ICGSE 13

ICGSE 26 CHASE 7

HICSS 15 OTM 3

ICSE 8 Global Sourcing 3

upon specific groups, societies and individuals are

found to be limited. Organizational concerns have

been at the forefront in terms of the level of analysis,

leaving much scope for consideration of groups and

individuals for future studies.

Figure 5: Number of locations used in GSE projects.

4.7 Location of GSE Projects

Figure 6 and Table 5 provide graphical and tabular

representations of the locations involved, whereas

Figure 5 depicts the spread of the number of

locations involved in GSE projects. Figure 5 shows

that most of the studies are focused on projects

involving two locations. A few studies also

mentioned regions rather than countries; we

considered them in the next section. In Figure 6,

countries and regions marked by darker shades are

0.36 %

1.46 %

1.82 %

7.30 %

21.53

67.15 %

0,00 20,00 40,00 60,00 80,00

9 locations

6 locations

5 locations

4 locations

3 locations

2 locations

TopicsandTreatmentsinGlobalSoftwareEngineeringResearch-ASystematicSnapshot

those most frequently involved in GSE. For ease of

analysis we grouped these countries into six

categories based upon the number of studies that cite

their involvement in global projects. Not

unexpectedly, the two countries reported as most

frequently involved in global software projects are

the USA and India. Countries including Germany,

Finland, China, the UK, Australia and Brazil are

ranked in the second group, closely followed by a

group comprising Sweden, Japan, Argentina, the

Netherlands, Spain, Canada and Switzerland. In the

next two categories lie the potentially upcoming and

emerging countries of Russia, Eastern European

countries such as Lithuania, Far Eastern countries

including Malaysia and Indonesia, and the

South/Central American countries of Chile and

Mexico. These representations give some insight

into the diversity of countries’ involvement in GSE

projects. Some of these regions are underrepresented

but this does not necessarily mean that these

locations are not involved in GSE; it could be that

these regions have simply not been considered in

recent studies. Researchers often seek industrial

contacts to validate their research outcomes and

gather feedback to improve their results, and often

times they rely on personal contacts in their national

industries. Our study also shows that the top seven

locations of GSE authors are the USA, Finland,

Germany, Spain, Brazil, India and Sweden. Apart

from Spain, which is thirteenth, all six other

countries are in the list of top ten locations involved

in GSE projects.

4.8 Inter-country Relationship Analysis

Figure 7 shows the results of our inter-country

network analysis. We used NodeXL, an extendable

tool kit used for data analysis and visualizations and

an add-in to Microsoft Excel spread sheet software

(Smith et al., 2009), to support our analysis. Table 6

lists the pairwise relationships with frequency

greater than one. (This constraint was imposed due

to space limitations; however, all the relationships

are shown in Figure 7.) It can be seen in Figure 7

and Table 6 that the most connected nodes are the

USA and India. Some studies explicitly mentioned

the collaborating locations whereas others only

specified the locations involved without clearly

stating which actively collaborated. For the latter

studies, we assumed pairwise relationships between

each location. For future studies we recommend that

authors clearly state the nature of each party’s

involvement. A few studies also mentioned regions

rather than countries; we used these as needed in

Figure 7.

5 THREATS TO VALIDITY

One of the main threats to the validity of our study is

the incomplete selection of primary studies or

missing relevant studies. There is a possibility that,

even though we followed a systematic process, we

may have missed some related work. In order to

mitigate this risk we formulated a wide variety of

search-terms. These terms were taken from related

SM/SLR studies and were updated based upon the

retrieved results. Initially, we ensured that at least

those SM/SLR studies were indeed retrieved using

the search terms drawn from each study. In the next

stage, we constructed a sample list of studies from

various ICGSE proceedings and ensured that the

search terms retrieved these studies as well. During

this process the search terms were continuously

updated until all sample studies were retrieved,

similar to the approach taken by (Jalali and Wohlin,

Figure 4: Bubble plot analysis.

ENASE2013-8thInternationalConferenceonEvaluationofNovelSoftwareApproachestoSoftwareEngineering

2010). A second validity threat arises due to

researcher bias during the classification process. In

order to reduce this threat, we carried out some

sample classifications collectively. Furthermore, the

lists of studies as classified by the first author were

validated by the senior researchers involved. A high

level of agreement was achieved, giving us

confidence that the classification process was

executed appropriately and consistently.

Table 5: Locations involved in GSE projects.

Country Frequency Country Frequency Country Frequency

US 238 Italy 11 Estonia 4

India 159 Norway 11 Philippines 4

Germany 57 Czech 10 Thailan

Finlan

54 Lithuania 10 Vietna

China 44 Israel 9 Korea 4

38 Malaysia 9 Costa Rica 3

Australia 32 Mexico 9 Colombia 3

Brazil 32 Senegal 9 Ecuado

Sweden 27 Singapore 9 Egypt 3

Japan 20 New Zealan

8 Greece 3

Argentina 19 Cambodia 7 Polan

Netherlands 19 France 7 Taiwan 3

Spain 18 Belgiu

6Romania 2

Canada 16 Chile 6 Slovakia 2

Switzerlan

16 Croatia 6 Turke

Ukraine 15 Hungar

6 Bangladesh 1

Russia 13 UAE 5 Pakistan 1

Denmar

11 Panama 5 South Africa 1

Irelan

11 Austria 4 Tunisia 1

Figure 6: Locations involved in GSE projects.

TopicsandTreatmentsinGlobalSoftwareEngineeringResearch-ASystematicSnapshot

Table 6: Inter-country relationships.

Location A Location B Frequency Location A Location B Frequency Location A Location B Frequency

India US 67 Singapore US 4 Australia Spain 2

China US 23 UK Ukraine 4 Australia Germany 2

Germany India 14 US Singapore 4 Belgium US 2

Brazil US 11 US Russia 4 Brazil UK 2

Australia US 10 US Israel 4 Cambodia Senegal 2

Europe US 10 China Japan 3 Cambodia India 2

UK US 10 Denmark India 3 Canada India 2

Finland India 8 E. Europe Finland 3 Canada Europe 2

Germany US 8 Finland Sweden 3 Europe Japan 2

Australia India 7 Finland Lithuania 3 Finland Japan 2

India Europe 7 Finland Baltic C. 3 Finland Brazil 2

UK India 7 Germany Russia 3 France Germany 2

US Argentina 7 Germany Brazil 3 Germany Czech 2

Finland US 6 India Senegal 3 India Switzerland 2

US Ukraine 6 India Japan 3 India Middle East 2

US Finland 6 India China 3 Ireland China 2

US Canada 6 Italy Switzerland 3 Lithuania US 2

Croatia Sweden 5 Japan India 3 Malaysia India 2

Czech Finland 5 Norway Finland 3 Netherlands Ukraine 2

Japan US 5 Spain Germany 3 Netherlands UK 2

Sweden Croatia 5 US Switzerland 3 New Zealand US 2

US Japan 5 US Sweden 3 Norway Sweden 2

US Ireland 5 US Spain 3 Norway Czech 2

W Europe India 5 US Senegal 3 Spain Lithuania 2

Brazil India 4 US Norway 3 Switzerland Vietnam 2

Finland Germany 4 US Mexico 3 Switzerland Ukraine 2

India Sweden 4 US Malaysia 3 US Taiwan 2

India Argentina 4 US Egypt 3 US Middle East 2

Netherlands US 4 US Denmark 3 US Cambodia 2

Netherlands India 4 Asia US 2 W Europe US 2

6 CONCLUSIONS

Through this study we have provided a current

snapshot of the GSE-related literature. We classified

275 empirical and non-empirical studies, published

between January 2011 and June 2012, into

predefined categories (see

http://tinyurl.com/GSE-

Papers). We examined the following characteristics:

GSE factors, research approaches, research methods,

level of analysis, and GSE project locations. The

GSE factors most frequently researched were related

to management and infrastructure using evaluative

approaches and taking an organizational perspective

as the level of analysis. Regarding research methods,

interviews, surveys, case studies and field studies are

the most commonly used. In relation to project

locations, the USA and India are the predominant

nations involved in global software projects. Inter-

country network analysis also shows that USA-India

collaboration is at the top followed by USA and

China. It will be interesting to carry out further

similar snapshot studies on an on-going basis to see

if or how these trends evolve. Similarly, studies

could be carried out retrospectively on previous

ENASE2013-8thInternationalConferenceonEvaluationofNovelSoftwareApproachestoSoftwareEngineering

Figure 7: Inter-country relationship analysis.

years’ research literature to enable comparisons with

this study. This study aims to provide a stepping

stone for further related studies.

It appears that, in general, existing solutions are

being applied in a GSE context, even though these

solutions may lack specific considerations needed

for GSE. For instance, aspects of non-functional

requirements and stage/phase-related issues are not

addressed separately in the current GSE literature.

Although the field of GSE research has grown

rapidly in terms of the number of studies conducted,

these studies are quite narrowly focused towards

exploratory research and the provision of

explanatory theories. Furthermore, in spite of GSE

providing a natural and potentially fruitful setting for

critical research, such work is yet to be conducted.

The current research focus is mainly directed to

organizational concerns, leaving much scope for

consideration of the needs of stakeholder groups and

individuals. The research is also skewed towards

projects having two locations showing a dearth of

studies relating to multiple locations and their

underlying complex relationships. Finally, there are

regions of the world that are not being currently

studied by researchers and it may be useful to

consider them in the future studies, particularly if the

dimensions of culture and their impact on GSE are

of interest.

7 FUTURE WORK

A notable omission in the current focus of work

relating to GSE is any sustained coverage of issues

to do with power and exploitation. While the human

factors tabulated in Table 2 above include some

focus on the factors of fear, trust, cooperation and

relationship, these are given relatively limited

attention. Again in Figure 4 there is a noted absence

of studies at an individual unit of analysis. There are

no studies giving personal narratives or biographies

– are the workers in GSE deliberately kept invisible?

Is this absence a function of the research methods

used, for instance, no examples of critical evaluative

work have been identified in this review? Or is it an

abrogation of our duties as academics to act in the

role of ‘critic and conscience of society’? Will the

future see more equal partnerships in sustainable

global ventures, or will there be a backlash against

crude models of global labour arbitrage? What risks

might that pose to a multi-billion dollar industry?

These issues warrant more attention by researchers,

although difficult to confront. In addition such

research will be challenging to design and conduct,

yet the absence of critical evaluative studies presents

a glaring gap in current GSE research.

REFERENCES

Alain, A., Pierre, B., Robert, D. & James, W. M. 2001.

Guide to the Software Engineering Body of Knowledge

- SWEBOK, IEEE Press, Piscataway, NJ, 2001.

Bailey, J., Budgen, D., Turner, M., Kitchenham, B.,

Brereton, P. & Linkman, S. Evidence relating to

Object-Oriented software design: A survey. In:

Empirical Software Engineering and Measurement,

2007. ESEM 2007. First International Symposium on,

20-21 Sept. 2007 2007.

TopicsandTreatmentsinGlobalSoftwareEngineeringResearch-ASystematicSnapshot

Clear, T. & MacDonell, S. G. 2011. Understanding

technology use in global virtual teams: Research

methodologies and methods. Inf. Softw. Technol., 53,

pp.994-1011

Da Silva, F. Q. B., Prikladnicki, R., França, A. C. C.,

Monteiro, C. V. F., Costa, C. & Rocha, R. 2011. An

evidence-based model of distributed software

development project management: results from a

systematic mapping study. Journal of Software

Maintenance and Evolution: Research and Practice,

pp.n/a-n/a

Dieste, O. & Padua, O. a. G. Developing Search Strategies

for Detecting Relevant Experiments for Systematic

Reviews. In: Empirical Software Engineering and

Measurement, 2007. ESEM 2007. First International

Symposium on, 20-21 Sept. 2007 2007.

Glass, R. L., Vessey, I. & Ramesh, V. 2002. Research in

software engineering: an analysis of the literature.

Information and Software Technology, 44, pp.491-506

Gregor, S. 2006. The Nature of Theory in Information

Systems. MIS Quarterly, 30, pp.611-642

Jalali, S. & Wohlin, C. Agile Practices in Global Software

Engineering - A Systematic Map. In: Global Software

Engineering (ICGSE), 2010 5th IEEE International

Conference on, 23-26 Aug. 2010 2010.

Kitchenham, B. & Charters, S. 2007. Guidelines for

performing Systematic Literature Reviews in Software

Engineering.

Mohd Fauzi, S. S., Bannerman, P. L. & Staples, M.

Software Configuration Management in Global

Software Development: A systematic map. In, 2010.

Morrison, J. & George, J. F. 1995. Exploring the software

engineering component in MIS research. Commun.

ACM, 38, pp.80-91

Myers, M. D. & Klein, H. K. 2011. A set of principles for

conducting critical research in information systems.

MIS Q., 35, pp.17-36

Orlikowski, W. J. & Baroudi, J. J. 1991. Studying

Information Technology in Organizations: Research

Approaches and Assumptions. Information Systems

Research, 2, pp.1-28

Petersen, K., Feldt, R., Mujtaba, S. & Mattsson, M.

Systematic mapping studies in software engineering.

In: Proceedings of the 12th international conference

on Evaluation and Assessment in Software

Engineering, 2008 Italy. British Computer Society.

Portillo-Rodríguez, J., Vizcaíno, A., Piattini, M. &

Beecham, S. 2012. Tools used in Global Software

Engineering: A systematic mapping review.

Information and Software Technology,

54, pp.663-685

Richardson, I., Casey, V., Mccaffery, F., Burton, J. &

Beecham, S. 2012. A Process Framework for Global

Software Engineering Teams. Information and

Software Technology,

Šmite, D., Wohlin, C., Feldt, R. & Gorschek, T. Reporting

empirical research in global software engineering: A

classification scheme. In, 2008.

Šmite, D., Wohlin, C., Gorschek, T. & Feldt, R. 2010.

Empirical evidence in global software engineering: a

systematic review. In Empirical Software Engineering,

15, pp.91-118

Smith, M. A., Shneiderman, B., Milic-Frayling, N.,

Rodrigues, E. M., Barash, V., Dunne, C., Capone, T.,

Perer, A. & Gleave, E. 2009. Analyzing (social media)

networks with NodeXL. Proceedings of the fourth

international conference on Communities and

technologies. University Park, PA, USA: ACM.

Steinmacher, I., Chaves, A. P. & Gerosa, M. A. 2012.

Awareness Support in Distributed Software

Development: A Systematic Review and Mapping of

the Literature. Computer Supported Cooperative

Work: CSCW: An International Journal, pp.1-46

ENASE2013-8thInternationalConferenceonEvaluationofNovelSoftwareApproachestoSoftwareEngineering