Using GitHub Cloud Service in Training Future IT Professionals: Local

Study

Olena G. Glazunova

, Valentyna I. Korolchuk

, Tetiana V. Voloshyna

Yevhenii M. Starychenko

and Oleksandra V. Parkhomenko

National University of Life and Environmental Sciences of Ukraine, 15 Heroiv Oborony Str., Kyiv, 03041, Ukraine

Keywords:

GitHub, Future IT Professionals, Collective Development.

Abstract:

In today’s IT industry, it is important to develop the ability of IT students to collaboratively develop software,

professional and personal skills. An effective method for developing such skills in future IT specialists is

to organize different types of educational projects related to different programming technologies during the

execution of mini projects, group and individual project assignments, term papers, academic training within

the academic disciplines. The paper summarizes the results of a pedagogical study involving 29 expert students

who study Computer Science and Software Engineering and used cloud service for GitHub collaborative IT

development projects. The research ﬁndings testify, the most effective characteristics of this service, according

to experts, identiﬁed the possibility of collaborative development of software (i1), the convenience of bug

tracking (i3) and the convenience of the code editor (i7). It offers examples and results of using GitHub cloud

service in the process of executing educational projects by future IT specialists.

1 INTRODUCTION

With the development of information technology (IT),

the approach to the organization of collaborative de-

velopment of software products is changing. Hence, it

is necessary to take into account the fact that future IT

specialists should be able to adapt instantly to new sit-

uations, make appropriate decisions and quickly solve

their tasks not only personally, but also while work-

ing as a team. In order for students of IT profession

to continue to hold leading positions in IT industry in

their professional activity, to meet the requirements of

customers and employers, it is necessary to develop

in them the ability to design and manage projects, to

work in a team, to develop skills to use cloud ser-

vices for project management and team development

of software products in the process of their academic

training at the university.

https://orcid.org/0000-0002-0136-4936

https://orcid.org/0000-0002-3145-8802

https://orcid.org/0000-0001-6020-5233

https://orcid.org/0000-0001-8608-5268

https://orcid.org/0000-0001-8773-0185

2 THEORETICAL BACKGROUND

Cloud software for team development of software

products allows users to collaborate on code, man-

age their versions, and more. Cloud services such as

GitHub, Bitbucked, GitLab, Phabricator, Beanstalk,

which were researched and described in the paper

(Korolchuk, 2019). become part of the cloud-oriented

scientiﬁc and educational environment of the univer-

sity if used on a regular basis (Glazunova and Shyshk-

ina, 2018). GitHub is the most popular code manage-

ment platform for software development, as it enables

future IT professionals to manage and collaborate on

their software development training projects.

GitHub is an online Git service that hosts Git

repositories and provides other features such as is-

sue tracking. GitHub has become the prominent plat-

form for hosting open source projects (Metz, 2015).

GitHub has been embraced by the software develop-

ment community as an important social platform for

managing software projects and to support collabo-

rative development (Feliciano et al., 2016). The most

important beneﬁt of using GitHub is not to support the

short-term priorities of a semester-long course, but,

rather, to encourage sustainable and well-documented

digital development, both of student projects and the

course itself (Beshero-Bondar and Parker, 2017).

448

Glazunova, O., Korolchuk, V., Voloshyna, T., Starychenko, Y. and Parkhomenko, O.

Using GitHub Cloud Service in Training Future IT Professionals: Local Study.

DOI: 10.5220/0012065200003431

In Proceedings of the 2nd Myroslav I. Zhaldak Symposium on Advances in Educational Technology (AET 2021), pages 448-461

ISBN: 978-989-758-662-0

 2023 by SCITEPRESS – Science and Technology Publications, Lda. Under CC license (CC BY-NC-ND 4.0)

When using GitHub in education, one has to think

about the purpose, what the goal is and then how the

features in GitHub can be used to reach this goal.

GitHub can be used in many different ways, but it

might not be applicable in all types of courses because

a certain amount of knowledge about Git is required

to be able to use the features (Gunnarsson et al.,

2017). Git and GitHub into data science workﬂows

is considered best practice, and provides thoughtful

advice on how to conceptualize the GitHub workﬂow

(Fiksel et al., 2019). Other work describes a GitHub

Education study that shows that using GitHub in the

classroom can lead to a much improved understand-

ing of students’ project management (GitHub Educa-

tion, 2020).

The most important skills employers seek in en-

gineering are creativity, teamwork and critical think-

ing. Kert

esz (Kert

esz, 2015) presents the results of

a collaborative learning experiment using GitHub in

lab work, where the focus was on students’ direct

interaction with each other’s learning process. De-

pending on how GitHub is implemented in learning

programming, students may rely on GitHub for ac-

tivities such as, submitting assignments, collaborat-

ing on group projects, and receiving feedback (Hsing

and Gennarelli, 2019). An immediate advantage is for

classes that have group projects. With GitHub Class-

room, instructors can easily assign groups of students

to teams and give each team their own GitHub repos-

itory within a GitHub Classroom. Students can then

use Git and GitHub to collaborate on a project, just

as they would in an academic or industry research

project. Because teachers can see each student’s com-

mit history, it is easy to see how each student con-

tributed to the project (Fiksel et al., 2019). A collabo-

rative tutorial assignment on the GitHub platform was

embedded in an undergraduate cybersecurity course.

Students were asked to create a tutorial that would

be combined with their peers’ tutorials to create a

course eBook. The tutorial topics were required to be

in the general domain of network security. With re-

gards to tutorial difﬁculty, students were told to target

an audience that had completed an introductory com-

puter networking course (Marquardson and Schuet-

zler, 2019).

Instead, this study was aimed at answering the fol-

lowing research questions: a) how to use the cloud

service to collectively develop GitHub for program-

ming training projects; b) which function-based as-

sessment indicators affect the effectiveness of the

cloud service for collective GitHub development;

c) how collective GitHub development inﬂuences the

formation of professional programming competence

of future information technology specialists. The mo-

tivation for this study was to demonstrate the effec-

tiveness of using the GitHub cloud service for student

programming projects.

The problem of the research stems from the need

to ﬁnd effective information and communication tech-

nologies for the organization and implementation of

various types of collaborative projects by future IT

specialists. Therefore, this study was conducted to

determine the characteristics of the GitHub cloud ser-

vice, which affect the effectiveness of the execution

of educational projects on programming in the pro-

cess of training IT students and in order to form pro-

fessional programming competence.

3 RESEARCH METHODOLOGY

3.1 General Background

In order to determine how effectively GitHub’s cloud

service enables students to carry out programmatic

learning projects, such as interacting with team mem-

bers, collectively working on code, sharing ideas,

and reviewing each other’s work during research,

students’ thoughts and impressions about using this

cloud service were gathered. A descriptive study uti-

lizing survey methodology was used as appropriate to

achieve the objectives of the study. This allowed the

researchers to gain a more detailed view of the stu-

dents regarding the use of the GitHub cloud service

in the execution of educational programming projects

using pedagogical observation and peer review meth-

ods. The study was conducted in two stages. At the

ﬁrst stage, the role of experts was performed by stu-

dents who evaluated the effectiveness of the GitHub

cloud service for the implementation of programming

training projects. The study of the ﬁrst stage was con-

ducted among the 3-rd year students of the Faculty

of Information Technologies of the National Univer-

sity of Life and Environmental Sciences of Ukraine

(NULES of Ukraine) during the second academic

term of 2019 (29 students).

Twenty-nine students of IT specialities performed

a collective mini-project in the process of studying

one informative module in the discipline “Object Ori-

ented programming”. The study of this academic dis-

cipline was preceded by the study of “Database Orga-

nization” discipline; “Development of a program sys-

tem for working with IT company management com-

puter systems databases” was selected as the project

theme. After completion of the technological practi-

cal training, an expert evaluation of the GitHub cloud

service was conducted by the students in the sec-

ond phase. To understand the attitude of students

Using GitHub Cloud Service in Training Future IT Professionals: Local Study

449

to the service for collaborative development, the fol-

lowing indicators of their evaluation from the point

of view of functionality were determined: (i1) possi-

bility of collaborative development of software; (i2)

ability to manage code versions; (i3) convenience of

bug tracking; (i4) ability to organize and plan team-

work; (i5) communication capability; (i6) the ability

to support platforms; (i7) the convenience of the code

editor; (i8) security and privacy; (i9) availability of

wiki pages. In order to evaluate the GitHub collabora-

tive development cloud service by speciﬁc indicators,

a survey was developed, which consisted of 9 ques-

tions, in which the experts evaluated the importance

of the indicators by assigning a ranking number.

In 2020, the second phase of the study was con-

ducted, during which the effectiveness of the use of

the GitHub cloud service was evaluated in the train-

ing of future IT professionals for the formation of pro-

fessional competence. During experimental research

among 3rd year students of the faculty there were two

samples of students: control (96 students) and experi-

mental groups (99 students). Student assessment was

conducted during the study of the subject ”Object-

Oriented Programming” for two semesters and tech-

nological practice (internship).

In the experimental study, students performed two

mini-projects in academic disciplines and one group

project during technological practical training. The

ﬁrst part of the study was to carry out the students’

programming project using the resources of the e-

learning course (ELC) of the academic discipline

combined with the cloud service for the collective de-

velopment of GitHub.

3.2 Data Analysis

The experts evaluated the signiﬁcance of the devel-

oped indicators by assigning them a ranking number.

The highest rated factor was assigned a rank of 1.

The level of agreement of experts’ opinions was de-

termined by the coefﬁcient of concordance. The con-

cordance coefﬁcient was applied to assess the degree

of consistency among experts, which was calculated

by the formula:

W =

12S

− n)

where

S = (

∑

i j

−

∑∑

i j

)

i j

– evaluations of the ranks of each object of exam-

ination, n – number of criteria evaluated, m – number

of experts who evaluated the service. To calculate it,

the sum of the assigned ranks and deviation squares of

the rank sums from the average sum for each indicator

were determined. The statistical signiﬁcance of the

coefﬁcient of concordance was checked against the

Pearson correlation criterion χ

= m(n − 1)W . Based

on the sums obtained, the sum of the converted ranks

was determined and the weight of each indicator was

calculated to the formula, where s

i j

= x

max

− x

i j

Student’s t-test and analysis of variance were used

to test the effectiveness of using the GitHub cloud

service to develop professional competence. One-

factor and two-factor analysis of variance with inter-

group and intragroup factors – mixed-model analysis

of variance (mixed-model ANOVA). Student’s T-test

allows to check the equality of mean values in two

samples and is calculated by the formula:

t =



− M



where M

and M

– mean value in control and exper-

imental groups; σ

and σ

– standard deviation; N

and N

– sample sizes.

The variance two-factor analysis allowed to esti-

mate the effect of two factors on different samples of

objects, and the one-factor one on the inﬂuence de-

pending on the evaluation stage.

The following conditions were considered to ob-

tain reliable results of analysis of variance:

1. In the analyzed groups, the values of the depen-

dent variable should be normally distributed. In

this case, it is assumed that the value of the de-

pendent variable has a normal distribution within

each group, relative to the levels of factors. How-

ever, the response values do not have to have this

distribution. Another weakening of the normality

requirements is the normality of the distribution

of model residues.

2. Homogeneity (homoskedasticity) of group vari-

ances. That is, the values of the dependent vari-

able in each group must be statistically equal.

4 IMPLEMENTATION

One of the important types of projects in the process

of IT specialists training is the projects on the collab-

orative development of software products, and there-

fore it is important to prepare students for the imple-

mentation of such projects since college times, to de-

velop in them the necessary professional and personal

skills, in particular the skills of shared software de-

velopments. When choosing cloud services for col-

laborative development of software products, the fol-

lowing issues should be taken into account: interop-

erability on the code, bug tracking, discussion of the

AET 2021 - Myroslav I. Zhaldak Symposium on Advances in Educational Technology

450

code with other team members, management of ver-

sions of code and integration of additional services,

availability of a repository, wiki and code editor, etc.

There arises a need to integrate additional services so

that the cloud services of software products collabora-

tive development could enable us to manage projects.

While implementing software development projects,

the students cannot be restricted by cloud services

only while organizing the teamwork; the future IT

specialists also need the services, which will allow

them to work together on the product code they plan

to develop.

Given the rate of change in the ﬁeld of IT, the

number of cloud services for teamwork is constantly

increasing, but assessing the functionality of such ser-

vices, they can be subdivided into two categories:

cloud services for project management (1) and team

development of the software product (2).

Recently, educators also started using GitHub as

a teaching tool for programming courses by hosting

code samples and managing student tasks, and orga-

nizing teamwork (Angulo and Aktunc, 2019). The

ability to use version control is a valuable skill for

computer science graduates to possess. Git is a well-

established, well received source version control sys-

tem for the software development community and be-

yond (Bonakdarian, 2017; Kelleher, 2014; Haaranen

and Lehtinen, 2015).

To complete the programming projects, experi-

mental group students were asked to combine ETC

resources in different academic disciplines with the

GitHub service. In to complete the programming

projects, students’ experimental group were asked to

combine ETC resources in different academic disci-

plines with the GitHub service. Morze and Glazunova

(Morze and Glazunova, 2013) proposed the struc-

tural features, the ratio of form and content of the

smart course elements and its properties: individ-

ual learning paths, content personiﬁcation, the use

of training elements with links to public information

resources, interactive training elements, multimedia,

communication and cooperation elements are sub-

stantiated. Teachers of relevant disciplines and tech-

nological practical training placed tasks of collective

projects in the ELC. The ETC, posted by the teacher,

contained the theoretical material (Book, Lesson re-

sources) and course terminology (Glossary resource),

lab sessions assignments (Assignment resource), and

the exchange of useful resources and ﬁles (Database

resource). At GitHub, student teams create their own

projects, in which they can further collaborate on

code writing, use the repository, perform branches, is-

sue releases, and communicate with each other while

completing study project tasks. The scheme of com-

bining Moodle resources with the GitHub cloud ser-

vice is presented in ﬁgure 1.

Figure 1: The scheme of combining Moodle resources with

the GitHub cloud service.

For project or lab assignments, student groups can

work on a public repository sharing code and ideas. It

also allows cross-team communication with multiple

teams working on a larger project as it happens in in-

dustry settings or teams exchanging ideas and review-

ing each other’s works. GitHub’s cloud service pro-

vides users with a user-friendly web interface to the

repository, user proﬁle tools, change tracking, mes-

saging, comments and online access, allowing the in-

structor and students to track the contributions of each

team member so that students can be held accountable

for their work. Thus, we can single out the following

features of the GitHub cloud service (ﬁgure 2), which

are important in the course of the implementation of

educational projects on programming:

• programming: code editor; code versions man-

agement; bug tracking; platform support; avail-

ability of wiki pages;

• collaborative development: joint software devel-

opment; teamwork planning and organization; es-

tablishing communication; security and privacy.

Precisely these features, inherent in GitHub tools,

make it possible to apply different types of educa-

tional projects related to different programming tech-

nologies. This cloud service was offered to students

for completing mini projects, for group and individ-

ual project assignments, term papers within academic

disciplines.

In order to determine the effectiveness of the cloud

service for GitHub collaborative development, 29 stu-

dents were surveyed on the above-mentioned evalua-

tion indicators after the implementation of the pro-

gramming educational projects. A questionnaire was

developed to ask students to assess the importance of

each of the indicators:

• i1. Possibility of collaborative development of

software;

• i2. Ability to manage code versions;

Using GitHub Cloud Service in Training Future IT Professionals: Local Study

451

Figure 2: Classiﬁcation of GitHub cloud service features.

• i3. Convenience of bug tracking;

• i4. Ability to organize and plan teamwork;

• i5. Communication capability;

• i6. Ability to support platforms;

• i7. Convenience of the code editor;

• i8. Security and privacy;

• i9. Availability of wiki pages.

Table 1 provides an assessment of the results of

determining the effectiveness of cloud service for

GitHub team development in the course of execut-

ing educational projects on programming by future IT

specialists. The highest-rated indicator was assigned

a rank of 1.

The concordance coefﬁcient W = 0.85 indi-

cates a high degree of convergence among experts.

Pearson correlation criterion was calculated to as-

sess the signiﬁcance of the concordance coefﬁcient

= m(n − 1)W . As the calculated one χ

(197.2)

is higher than the table value (15.5) for the number of

degrees of freedom K = n − 1 = 9 − 1 = 8 and at a

given level of importance α = 0.05, we may conclude

that the obtained coefﬁcient of concordance of 0.85 is

not accidental, and therefore the results obtained are

statistically signiﬁcant.

Based on the obtained rank sum the weights of

the indicators considered were calculated. The survey

matrix was transformed into a matrix of transformed

ranks according to the formula, where s

i j

= x

max

−x

i j

in which x

max

= 9 and the weight of each indicator

was calculated.

The analysis of the signiﬁcance of the factors stud-

ied revealed that the following indicators were noted

by the students as being the most signiﬁcant ones: the

possibility of collaborative development of software,

the convenience of bug tracking, the convenience of

the code editor, and the ability to manage code ver-

sions when completing educational projects on pro-

gramming.

In the second stage of the study, students worked

on tasks of various projects using the cloud service

GitHub. It was proposed to carry out collective mono-

projects during the study of professional disciplines or

course work within such disciplines, which will allow

the formation of future IT professionals professional

competencies and soft skills using services for collec-

tive IT development for inverted learning (Glazunova

et al., 2022).

While studying “Object Oriented Programming”

academic disciplines students were asked to com-

plete mini projects using a cloud service to collab-

oratively develop GitHub software. The purpose of

such projects was to develop professional competen-

cies and personal effectiveness skills in future IT spe-

cialists Students worked in teams of 4-5 people. In

each team a leader was identiﬁed, he distributed the

tasks among the participants of the collective project.

The task of “Database Organization” organization

mini projects was to design a relational model of

databases for the future automated system (according

to the topic chosen by students); constructing class

diagrams and developing a system using a class com-

position.

AET 2021 - Myroslav I. Zhaldak Symposium on Advances in Educational Technology

452

Table 1: Assessment of the results of determining the effectiveness of cloud service for GitHub team development in the

course of executing educational projects on programming by future IT specialists.

Indicator Rank sum S

Concordance

coefﬁcient

Pearson criterion Sum of

converted ranks

Indicator

weightpredictive table

i1 35 12100

0.85 197.2 15.5

226 0.22

i2 121 576 140 0.13

i3 62 6889 199 0.19

i4 147 4 114 0.11

i5 173 784 88 0.08

i6 248 10609 13 0.01

i7 96 2401 165 0.16

i8 227 6724 34 0.03

i9 196 2601 65 0.06

∑

1305 42688 1044 1

Within the framework of mastering the ”Ob-

ject Oriented Programming” academic discipline, the

students were offered to implement an educational

project entitled “Development of a program system

for working with IT company management com-

puter systems databases”, the objective of which was

to review and analyze modern design technologies;

to develop and use software standards for common

computer-driven control systems; to develop software

structures for the computerized management system

and UML diagrams of design entities; to develop

a graphical interface for computer control system

software; programming and debugging using object-

oriented programming techniques; testing and anal-

ysis of the performance of the developed computer

management software; reporting on the performance

of computer-based management systems; develop-

ment of a set of standard documents to support the

developed computer management software.

Within the framework of the project and tech-

nological practical training, students performed a

collaborative project using the GitHub cloud ser-

vice aimed at the development of their professional

and personal competencies, namely: improvement of

practical skills in software development and design

using modern approaches and tools for ﬂexible soft-

ware development, development of teamwork skills in

students, which are in demand on the modern IT labor

market. The task of the team project was to develop

software with web interface and relational database

using HTML, CSS, JavaScript, MySQL, PHP tech-

nologies. Work on the educational project was carried

out in line with the principles of Agile ﬂexible devel-

opment and Scrum methodology, which provides an

incremental and iterative approach and speciﬁc roles

of the participants in the development of the collabo-

rative project. In the course of collaborative IT devel-

opment using GitHub, future IT specialists kept the

educational project code and necessary documenta-

tion in the public domain. In addition, a version con-

trol system was used to provide integrity and multi-

user access.

5 EXPERIMENTAL RESULTS

To assess the effectiveness of the use of GitHub, two

samples of students were selected: a control group

(Control) without the use of a joint development ser-

vice in the educational process and an experimen-

tal one using GitHub (Experiment). The comparison

will be made in three stages of studying the discipline

”Programming”. Assessments were conducted in the

following sequence: exam for the ﬁrst semester (Exa-

men1), results for the second semester (Examen2) and

internship (EducPractic). Accordingly, the study put

forward the following null hypotheses, the deviation

of which will conﬁrm the effectiveness of the use of

the cloud service GitHub for the formation of profes-

sional competence: 1) the average score in the control

and experimental groups does not differ; 2) the differ-

ence in the average score at different stages of eval-

uation is statistically insigniﬁcant; 3) the difference

in the average score by groups (samples) at different

stages of evaluation is statistically insigniﬁcant.

To test the ﬁrst hypothesis, Student’s t-test or

its non-parametric analogue, the Mann-Whitney-

Wilcoxon test, will be used. To check others – anal-

ysis of variance. In particular, for the second hy-

pothesis, one-factor analysis of variance (if necessary,

non-parametric Kruskal-Wallis test) and for the third,

two-way analysis of variance with intergroup and in-

tragroup factors – mixed-model analysis of variance

(mixed-model ANOVA).

Before starting the statistical veriﬁcation, the

power analysis should be used to determine the level

of effect for the given methods that provide sample

Using GitHub Cloud Service in Training Future IT Professionals: Local Study

453

sizes. Table 2 shows the sample sizes in different sec-

tions. A signiﬁcance level of 0.05 and a power of 80%

were also selected for power analysis.

Table 2: The size of student samples at each stage of the

study.

Evaluation stage

Group

Sum

Control Experimental

Examen1 96 99 195

Examen2 96 99 195

EducPractic 96 99 195

Sum 288 297 -

To assess the possibility of neglecting the normal-

ity requirement, the effect level for Student’s t-test

was determined by the ﬁrst hypothesis for samples of

experimental and control groups, which are 288 and

297, respectively. The results of the calculations are

presented in listing (ﬁgure 3).

Figure 3: Evaluation of the effect for the Student’s t-test.

As ﬁgure 3 of the power analysis shows, the effect

level is 0.232 (d value), which indicates the possibility

of determining small effects (Cohen, 1988), accord-

ingly, the requirement for data distribution according

to the normal distribution law can be neglected.

To analyze the capacities for small groups, a cal-

culation was made for pairwise comparisons of the

obtained scores in terms of individual types of evalu-

ation (sizes of groups 96 and 99), which is presented

in listing (ﬁgure 4).

Figure 4: Evaluation of the effect for pairwise comparisons

of scores in terms of individual types of evaluation.

The obtained value of the effect (0.403) corre-

sponds to the average theoretical level of the effect,

therefore, when choosing a statistical procedure, it is

advisable to consider the law of data distribution in

the samples.

For the second hypothesis, where the test method

should be a one-way analysis of variance with a group

size of 195, the calculated effect is shown in listing

(ﬁgure 5).

Figure 5: Effect evaluation for analysis of variance.

The obtained value (0.129) is close to the theoreti-

cal value of the small effect (0.1) but slightly exceeds

it. Therefore, when establishing the inconsistency of

the distribution with the normal law, we additionally

use nonparametric methods.

Since there is no statistical procedure for de-

termining the level of effect in the power analysis

for two-factor analysis with an unbalanced design

(groups of different sizes), this analysis was not per-

formed. The obtained average scores in the sections

of the groups and evaluation by the results of the sec-

ond stage of the experiment are presented in table 3

and in ﬁgure 6.

Table 3: The average performance of students in the control

and experimental groups in terms of stages of assessment.

Evaluation

stage

Group By types of

evaluationControl Experimental

Examen1 73.1 75.9 74.5

Examen2 76.2 79.9 78

EducPractic 74.3 82.2 78.2

By groups 74.6 79.3 -

According to the summary data, the difference be-

tween the overall scores in the groups is 4.7 points. At

the same time, if we evaluate the pairwise differences

in grades for different types of assessment, the biggest

difference was in the success of students in the results

of internship – 7.9 points.

Analyzing the data presented in ﬁgure 3, we see

the difference in the medians, as well as the distribu-

tion of scores – the experimental group shows the best

results for both general and stages (ﬁgure 7).

In this case, if the experimental group is character-

ized by an increase in scores in stages: Examen1 →

Examen2 → EducPractic; then in the control group

after the growth of average scores, the average scores

of internship results are lower than for the exam in the

AET 2021 - Myroslav I. Zhaldak Symposium on Advances in Educational Technology

454

Figure 6: Average student performance based on the results

of three stages of assessment.

Figure 7: The success of students of control group and ex-

perimental group in terms of assessment stages.

2nd semester.

Comparing estimates by type of control, there is

an increase in scores by stages (ﬁgure 8). At the same

time, a signiﬁcant change occurred between exams in

different semesters (3.5 points).

The difference between such related stages as the

exam in the 2nd semester and internship is insigniﬁ-

cant (0.2 points), which is caused as noted above by

the deterioration of grades for internship in the control

group.

For the ﬁnal choice of methods for estimating sta-

tistical hypotheses, the samples were tested for dis-

tribution normality using the Shapiro-Wilk test. The

obtained results are presented in ﬁgure 9.

According to the data obtained from the Shapiro-

Wilk tests, the null hypothesis about the normal-

ity of the distribution laws of the control (p-value

= 1.382·10

−15

) and experimental groups (p-value =

Figure 8: Student performance in terms of assessment

stages.

Figure 9: The results of checking the samples (control

group and experimental group) for the normality of the dis-

tribution.

3.082·10

−11

) was rejected, which also illustrated the

graphs of the empirical and the theoretical distribution

density function.

Similarly, the veriﬁcation of the obtained data was

carried out by types of evaluation, which is presented

in ﬁgure 10.

The obtained results (p-values: 3.536·10

−10

7.687e·10

−11

and 1.741·10

−11

) indicate that these

samples are not subject to the normal distribution law.

Because analysis of variance is used to assess the sta-

tistical signiﬁcance of differences, a group variance

test (Barlett test) and an emission test (Bonferroni

Using GitHub Cloud Service in Training Future IT Professionals: Local Study

455

Figure 10: The results of checking the data on the normality

of the distribution by type of assessment.

test) were performed.

To conduct the Barlett test, the null hypothesis was

put forward that the variances between the groups are

equal. The estimate of the variance homogeneity in

the groups for different types of estimation is shown

in listing (ﬁgure 11).

Figure 11: Estimation of variance homogeneity for different

types of estimation.

The obtained value of p-value = 0.1359 is greater

than the signiﬁcance level – 0.05, so there is no reason

to reject the null hypothesis of the test on the equality

of variances in groups. The data obtained in this way

indicate that the variance data are statistically differ-

ent. Similarly, the statistical equality of variances for

the control and experimental groups of the third hy-

pothesis, which is presented in listing (ﬁgure 12), was

estimated.

Figure 12: Estimation of dispersion homogeneity for con-

trol group and experimental group.

As in the previous case, we decide to reject the

null hypothesis of equality of variances.

The Bonferroni test listings (ﬁgure 13) show no

emissions for both types of analysis of variance. Ac-

cordingly, the original data meet these two require-

ments for analysis of variance.

Figure 13: Bonferroni test.

Therefore, considering the obtained estimates of

the normality of the data distribution and the results

of power analysis, we choose the following methods

for testing hypotheses:

• Hypothesis 1: to check the equality of the average

overall scores – Student’s t-test; for comparisons

of average scores in groups for individual types of

evaluation – Mann-Whitney test, as in contrast to

the overall estimates, the size of the effect is esti-

mated as average, which does not allow to ignore

the normality of the distribution.

• Hypothesis 2: despite the fact that the size of

the effect is close to small, in addition to one-

way analysis of variance, the results were veriﬁed

through the Kruskal-Wallis test.

• Hypothesis 3: since it is impossible to perform

a power analysis for two-factor analysis of vari-

ance with an unbalanced design, the requirement

of normality was neglected due to the large sam-

ples in each group (Zar, 1996). Moreover, tests

for the homogeneity of variances and emissions

indicated the possibility of analysis of variance.

After analysis and selection of the above methods,

the hypotheses put forward in the study were tested.

AET 2021 - Myroslav I. Zhaldak Symposium on Advances in Educational Technology

456

According to a preliminary analysis, the differ-

ence between the overall scores in the groups is 4.7

points. To assess whether this difference in scores

is statistically signiﬁcant (Hypothesis 1), we chose in

the previous stages we chose Student’s t-test. Prior to

the test, a check was made for the Welch amendment

to be considered for samples with different variances

(ﬁgure 14).

Figure 14: Test for equality of variances in samples.

As you can see from the data presented in ﬁg-

ure 14, the probability of obtaining an error of the ﬁrst

kind is 78.6% with a permissible 5%, to reject the null

hypothesis. Therefore, the variances are statistically

equal and the Welch correction is not required. The

evaluation of the t-test for the overall averages in the

two groups is presented in listing (ﬁgure 15).

Figure 15: Checking the equality of average overall scores.

According to the obtained results, the actual value

of the criterion – t

= 4.25 exceeds the critical – t

r =

1.967 for a given level of signiﬁcance (0.05), which

is necessary to reject the null hypothesis of equality

of the two averages. Therefore, we can conclude that

the difference between the mean scores between the

control and experimental groups (4.7 points) is statis-

tically signiﬁcant. In this case, with a probability of

95%, this difference will be from 2.6 to 7.0 points.

Accordingly, the null hypothesis is rejected.

As noted earlier, a nonparametric Mann-Whitney-

Wilcoxon test, which is used for samples without a

normal distribution, was calculated to conﬁrm the dif-

ference in scores between samples at different stages

of the assessment (ﬁgure 16).

Figure 16: Calculation of the nonparametric Mann-

Whitney-Wilcoxon test.

The obtained test results indicate that a statisti-

cally signiﬁcant difference between the averages in

the groups is observed only for internship (7.9 points),

which is indicated by the value of p-value which is

less than the level of signiﬁcance.

However, it should be noted that in addition to the

fact that this test, like all non-parametric has less ac-

curacy, the use of pairwise comparisons with the pos-

sibility of analysis of variance is not statistically ef-

fective. Therefore, this analysis can be considered as

additional in a more efﬁcient analysis of variance.

As noted in the previous analysis, the difference

between the exams was signiﬁcant in contrast to the

exam (2 semester) and internship. To test the statis-

tical signiﬁcance of the differences in the mean score

at different stages (Hypothesis 2), the signiﬁcance of

the differences was estimated (ﬁgure 17).

Figure 17: Assessing the signiﬁcance of differences.

The F-test indicates that there are differences be-

tween the groups (p < 0.05). As noted at the stage

of selection of research methods, an additional anal-

ysis was performed according to the non-parametric

Kruskal-Wallis test (ﬁgure 18).

The results of the Kruskal-Wallis test conﬁrmed

the results of one-way analysis of variance.

Using GitHub Cloud Service in Training Future IT Professionals: Local Study

457

Figure 18: Analysis by the nonparametric Kruskal-Wallis

test.

Pairwise comparisons using the Tukey test (ﬁg-

ure 19) and graphical data (ﬁgure 20) concluded

that the difference between the exams (Examen2-

Examen1 – p-value = 0.029) and the exam for

1 semester was statistically signiﬁcant. Internship

(Examen1-EducPractic – p-value = 0.022) and is -3.7

and 3.5, respectively. The difference between the sec-

ond semester and the internship is not statistically sig-

niﬁcant (p-value = 0.994 and intersects on the graph

with a vertical January) and is within the statistical

error.

Figure 19: Conducting pairwise comparisons using the

Tukey test.

Figure 20: Graphical display of pairwise comparisons.

To check the statistical signiﬁcance of the differ-

ence in the mean score by groups (ﬁgure 21) at differ-

ent stages of the assessment (Hypothesis 3) was used

two-factor analysis of variance with intergroup and

intragroup factors – mixed analysis model (ANOVA).

In our study, the intergroup factor was the distribution

of groups of students relative to the control and exper-

imental, and intragroup – the stages of evaluation.

Figure 21: Plot group means and conﬁdence intervals.

The constructed variance model for two-factor

analysis of variance is presented in listing (ﬁgure 22).

Figure 22: Estimation of a two-factor variance model.

The calculated value of the Fisher criterion, based

on the mean squares of the deviations within and be-

tween groups and the corresponding degrees of free-

dom for each of the factors is:

• For groups:

νBG = m − −1 = 2 − 1 = 1

νW G = n − −m = 288 − −2 = 286

where m is the number of factor levels (groups),

n is the number of observations (students)

Accordingly, according to tables F of Fisher’s

test at a signiﬁcance level of 0.05, the empirical

(theoretical, critical) value is F{0.05; 1; 286} =

3.847.

• For types of assessment:

νBG = m − −1 = 3 − 1 = 2

νW G = n − −m = 195 − −3 = 192

AET 2021 - Myroslav I. Zhaldak Symposium on Advances in Educational Technology

458

Accordingly, according to tables F of Fisher’s

test at a signiﬁcance level of 0.05, the empirical

(theoretical, critical) value is F{0.05; 2; 192} =

3.042.

• For stages and interaction of groups and stages

νBG = (m

–1) ∗ (m

–1) = (2 − 1) ∗ (3 − 1) = 2

νW G = (m

+ m

− 1) ∗ (n − 1) = (2 + 3 − 1) +

(

99+96

− 1) = 386,

where m

, m

is the number of levels for inter-

group (group) and intragroup (stages) factors,

n is the number of observations (students) in

each sample divided into groups and stages –

F{0.05; 2; 386} = 3.847.

According to the results of the study, the depen-

dence of the averages for groups and types of assess-

ment is statistically signiﬁcant, as indicated by the

value of the actual F statistics, exceeding the critical

values found: 9.247>3.890 and 9.257>3.042 (or ac-

cording to p-value – 0.0026 and 0.00011, respectively,

which is much less than the signiﬁcance level of 0.05).

That is, two-factor analysis of variance proved the

preliminary results on the deviation of 1 and the sec-

ond null hypotheses.

As for the interaction of groups and stages, is the

question in assessing the statistical signiﬁcance of dif-

ferences by groups at different stages, the value of

the obtained Fisher statistics is more than critical –

3.959>3,847 (p-value = 0.019871) indicates the pres-

ence of an additional effect from the interaction. That

is, there is a signiﬁcant difference in the average of

individual groups by type of assessment. For the ob-

tained model we will make multiple comparisons ac-

cording to the Tukey test:

• The group comparison (ﬁgure 23) conﬁrmed the

t-test data on the statistical signiﬁcance of the dif-

ference in the rating score for the control and ex-

perimental groups, which is 4.7 points.

Figure 23: Multiple group comparisons.

• From the obtained comparisons (ﬁgure 24) of the

averages by stages (types of evaluation) for all

groups, we see that the signiﬁcant difference in

the average evaluations corresponds to the data of

one-factor analysis for the second hypothesis.

• Pairwise comparisons of the difference between

the mean scores for the groups divided by both

factors are signiﬁcant for most comparisons (ﬁg-

ure 25).

Figure 24: Multiple comparisons by type of assessment.

Figure 25: Multiple comparisons by groups and types of

assessment.

Comparisons between the same types of evalu-

ations between the control and experimental groups

showed:

• The difference in the average scores on the exam

for the ﬁrst semester (Control Examen1 – Exper-

imental Examen1) is statistically insigniﬁcant (p-

value = 0.7052>0.05);

• The difference in the average scores on the exam

for the second semester (Control Examen2 – Ex-

perimental Examen2) is statistically insigniﬁcant

(p-value = 0.4087>0.05);

• The difference in the average scores on in-

ternship (Control EducPractic – Experimental

EducPractic) is statistically signiﬁcant (p-value =

0.0007<0.05) and is 7.8 points.

These results conﬁrmed the results of the Mann-

Whitney-Wilcoxon test that the effectiveness of the

implementation of experimental methods affected the

results of internship, during which students actively

used the cloud service GitHub to implement collec-

tive projects.

Assessing separately the dynamics of differences

in each of the groups by stages, we see that in the con-

trol group, each change in the average score between

different types of assessment is insigniﬁcant, which

indicates the absence of any dynamics in traditional

learning: “Control Examen1 – Control Examen2” –

p-value = 0.2034, “Control EducPractic – Control Ex-

amen2” – p-value = 0.7128, “Control EducPractic –

Control Examen1” – p-value = 0.9556. In contrast,

Using GitHub Cloud Service in Training Future IT Professionals: Local Study

459

for the experimental group where such changes were

statistically signiﬁcant.

6 CONCLUSIONS

The educational projects on programming are an ef-

fective method for shaping the professional and per-

sonal competencies of future IT specialists. To work

on educational projects, you should use modern cloud

services for collaborative IT development, such as

GitHub. The most effective features of this service

are the possibility of collaborative development of

software (i1), the convenience of bug tracking (i3),

and the convenience of the code editor (i7), which

are determined by the statistical processing of student

peer review. Other features of this service that have

also been explored include the ability to manage code

versions (i2), the ability to organize and plan team-

work (i4), the ability to communicate (i5), the ability

to support platforms (i6), security and privacy (i8),

availability of wiki pages (i9).

The GitHub cloud service can be applied to com-

plete mini projects, group or individual work, term

papers, or during the academic training of the stu-

dents. The examples of using GitHub discussed in

the paper show that the speciﬁc features of this service

completely satisfy the needs of students of IT profes-

sion in the implementation of the tasks of educational

projects on programming. And affects the formation

of professional competence of future specialists in in-

formation technology, as evidenced by the results of

the study. The obtained results of the experimental

group are higher by 5.93% than in the control group.

Crucial to the formation of professional competence

are the use of cloud service in the implementation of

collective projects for software development during

training.

Further research can be aimed at theoretical sub-

stantiation and development of methods of ﬂexible

training of future information technology specialists

using services for joint software development in the

implementation of educational projects, as the tech-

nology of agile learning is closest to real conditions

in software development.

REFERENCES

Angulo, M. A. and Aktunc, O. (2019). Using GitHub

as a Teaching Tool for Programming Courses. In

2018 Gulf Southwest Section Conference, Austin, TX.

https://peer.asee.org/31594.

Beshero-Bondar, E. E. and Parker, R. J. (2017). A GitHub

Garage for a Digital Humanities Course. In Fee, S. B.,

Holland-Minkley, A. M., and Lombardi, T. E., edi-

tors, New Directions for Computing Education: Em-

bedding Computing Across Disciplines, pages 259–

276. Springer International Publishing, Cham. https:

//doi.org/10.1007/978-3-319-54226-3 15.

Bonakdarian, E. (2017). Pushing Git & GitHub in Under-

graduate Computer Science Classes. J. Comput. Sci.

Coll., 32(3):119–125.

Cohen, J. (1988). Statistical Power Analysis for the Be-

havioral Sciences. Lawrence Erlbaum Associates,

2 edition. https://www.utstat.toronto.edu/

∼

brunner/

oldclass/378f16/readings/CohenPower.pdf.

Feliciano, J., Storey, M.-A., and Zagalsky, A. (2016). Stu-

dent Experiences Using GitHub in Software Engi-

neering Courses: A Case Study. In Proceedings of

the 38th International Conference on Software Engi-

neering Companion, ICSE ’16, page 422–431, New

York, NY, USA. Association for Computing Machin-

ery. https://doi.org/10.1145/2889160.2889195.

Fiksel, J., Jager, L. R., Hardin, J. S., and Taub, M. A.

(2019). Using GitHub Classroom To Teach Statis-

tics. Journal of Statistics Education, 27(2):110–119.

https://doi.org/10.1080/10691898.2019.1617089.

GitHub Education (2020). 2020 GitHub Education Class-

room Report: Insights from the next generation of

software developers. https://education.github.com/

classroom-report/2020.

Glazunova, O. G., Korolchuk, V. I., Parhomenko, O. V.,

Voloshyna, T. V., Morze, N. V., and Smyrnova-

Trybulska, E. M. (2022). Methodology for using

Cloud-oriented Environment for Flipped Learning of

the Future IT Specialists. In Semerikov, S., Os-

adchyi, V., and Kuzminska, O., editors, Proceed-

ings of the 1st Symposium on Advances in Educa-

tional Technology - Volume 1: AET, pages 445–

460. INSTICC, SciTePress. https://doi.org/10.5220/

0010925100003364.

Glazunova, O. G. and Shyshkina, M. (2018). The Con-

cept, Principles of Design and Implementation of

the University Cloud-based Learning and Research

Environment. In Ermolayev, V., Su

arez-Figueroa,

M. C., Yakovyna, V., Kharchenko, V. S., Kobets,

V., Kravtsov, H., Peschanenko, V. S., Prytula, Y.,

Nikitchenko, M. S., and Spivakovsky, A., editors, Pro-

ceedings of the 14th International Conference on ICT

in Education, Research and Industrial Applications.

Integration, Harmonization and Knowledge Transfer.

Volume II: Workshops, Kyiv, Ukraine, May 14-17,

2018, volume 2104 of CEUR Workshop Proceedings,

pages 332–347. CEUR-WS.org. https://ceur-ws.org/

Vol-2104/paper 158.pdf.

Gunnarsson, S., Larsson, P., M

ansson, S., M

artensson,

E., and S

onnerup, J. (2017). Enhancing Stu-

dent Engagement Using GitHub as an Educational

Tool. In Introduction to Teaching and Learning

in Higher Education. Genombrottet, Lunds tekniska

ogskola. https://lucris.lub.lu.se/ws/ﬁles/27854342/

group1 github ﬁnal.pdf.

Haaranen, L. and Lehtinen, T. (2015). Teaching Git on

AET 2021 - Myroslav I. Zhaldak Symposium on Advances in Educational Technology

460

the Side: Version Control System as a Course Plat-

form. In Proceedings of the 2015 ACM Conference

on Innovation and Technology in Computer Science

Education, ITiCSE ’15, page 87–92, New York, NY,

USA. Association for Computing Machinery. https:

//doi.org/10.1145/2729094.2742608.

Hsing, C. and Gennarelli, V. (2019). Using GitHub in the

Classroom Predicts Student Learning Outcomes and

Classroom Experiences: Findings from a Survey of

Students and Teachers. In Proceedings of the 50th

ACM Technical Symposium on Computer Science Ed-

ucation, SIGCSE ’19, page 672–678, New York, NY,

USA. Association for Computing Machinery. https:

//doi.org/10.1145/3287324.3287460.

Kelleher, J. (2014). Employing git in the classroom. In

2014 World Congress on Computer Applications and

Information Systems (WCCAIS), pages 1–4. https://

doi.org/10.1109/WCCAIS.2014.6916568.

Kert

esz, C.-Z. (2015). Using GitHub in the classroom -

a collaborative learning experience. In 2015 IEEE

21st International Symposium for Design and Tech-

nology in Electronic Packaging (SIITME), pages 381–

386. https://doi.org/10.1109/SIITME.2015.7342358.

Korolchuk, V. (2019). Cloud services for collective projects

preparation processes of future IT-professionals:

Analysis and selection criteria. New pedagogi-

cal thought, 100(4):46–51. https://doi.org/10.37026/

2520-6427-2019-100-4-46-51.

Marquardson, J. and Schuetzler, R. M. (2019). Teach-

ing Tip: Learning by Teaching through Collabo-

rative Tutorial Creation: Experience using GitHub

and AsciiDoc. Journal of Information Systems Ed-

ucation, 30(1):10–18. https://jise.org/Volume30/n1/

JISEv30n1p10.html.

Metz, C. (2015). How Github Con-

quered Google, Microsoft, and Every-

one Else. https://www.wired.com/2015/03/

github-conquered-google-microsoft-everyone-else/.

Morze, N. V. and Glazunova, O. G. (2013). What Should

be E-Learning Course for Smart Education. In

Ermolayev, V., Mayr, H. C., Nikitchenko, M. S.,

Spivakovsky, A., Zholtkevych, G., Zavileysky, M.,

Kravtsov, H., Kobets, V., and Peschanenko, V. S.,

editors, Proceedings of the 9th International Con-

ference on ICT in Education, Research and Indus-

trial Applications: Integration, Harmonization and

Knowledge Transfer, Kherson, Ukraine, June 19-22,

2013, volume 1000 of CEUR Workshop Proceedings,

pages 411–423. CEUR-WS.org. https://ceur-ws.org/

Vol-1000/ICTERI-2013-p-411-423-MRDL.pdf.

Zar, J. H. (1996). Biostatistical Analysis. Prentice Hall,

Upper Saddle River, 3 edition.

Using GitHub Cloud Service in Training Future IT Professionals: Local Study

461