Towards Improving Experimentation in Software Engineering

OliveiraJr, Edson; Furtado, Viviane R.; Vignando, Henrique; Luz, Carlos Danilo; Cordeiro, André F. R.; Steinmacher, Igor; Zorzo, Avelino F.

doi:10.1145/3474624.3477073

Cited by 7 publications

(2 citation statements)

References 18 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…As a way of defining the experimental methods, we identified the use of the approach with a certain objective of the study, selecting the variables to be observed and the form of control to be adopted. This information is corroborated in [14].…”

Section: Introductionsupporting

confidence: 63%

Evaluation and Use of a Student-Centered Syllabus for the Software Process Subject in a Postgraduate Course: A Quasi-Experiment

QUARESMA

Oliveira

2022

Education Sciences

View full text Add to dashboard Cite

Background: The software-development process is considered a knowledge area in software engineering that allows the performance of activities for the inception of a functional software. In a globalized software context, where companies and organizations have their data and information controlled by applications, it can be seen that the form of construction, that is, the path for which the software is built, has relevance for users. Problem Analysis: Within this reality, it is possible to observe that over the years, scientific research has identified competence deficits for professionals trained in software engineering and more specifically in the software-development processes. Purpose: This work aims to develop and apply a syllabus for the software-development process, with student-centered learning strategies, as it allows for centrality of the teaching and learning process for the student. Methods: The strategies were selected through a literature review, which informed which ones were adopted to intervene in the traditional teaching process in software. Other methodological procedures adopted to identification of the necessary requirements for the development of the syllabus were (i) a survey with the identification and perception of professionals and professors on the research theme, (ii) equivalence mapping of the main documents related to the theme of research at an educational level, and (iii) the identification of how to develop a syllabus. After gathering the requirements, the syllabus was designed and described in each of its parts based on the specialized literature. The evaluation of the syllabus took place in two stages: (i) an expert panel, where experts in the software-development process and student-centered learning were selected, who contributed points to be adjusted for the use of the syllabus, and (ii) the experiment, which was the use of the syllabus in comparison with the traditional approach (use of theoretical classes, exercise list, and written test), through the control and experimental groups. Results: Data generated by using the syllabus in the experiment, as well as data from the control group, were analyzed using the two-tailed Student-t technique. The results achieved in the work demonstrate that there was a considerable gain in learning compared to the control group. Another result point of the syllabus was that through the project strategies, students had contact with a progression of knowledge related to hypothetical cases that simulated reality, which students reported was very good for their knowledge. Conclusions: The use of student-centered strategies can bring benefits in the learning of software-development processes, as it allows a practical application of real scenarios observed in software-development companies.

show abstract

Section: Introductionsupporting

confidence: 63%

Evaluation and Use of a Student-Centered Syllabus for the Software Process Subject in a Postgraduate Course: A Quasi-Experiment

QUARESMA

Oliveira

2022

Education Sciences

View full text Add to dashboard Cite

show abstract

“…Even with advances observed in the ESE area [OliveiraJr et al 2021], there are still relevant issues to be addressed, such as, improving the level of reproducibility [González-Barahona and Robles 2012, Baker 2016, Anchundia et al 2020, Nelson et al 2021] and lack of sharing of experimental artifacts [Timperley et al 2021, Damasceno et al 2021, which prevent the software engineering evolution as a formal scientific discipline. It is understood that such an ESE evolution might be reached by increasing the formalization of the experiments carried out straightforwardly based on open science practices applied to SE [Mendez et al 2020].…”

Section: Introductionmentioning

confidence: 99%

Towards an Open Science-Based Framework for Software Engineering Controlled (Quasi-)Experiments

Cordeiro¹,

OliveiraJr²,

Capretz³

2022

Anais Do XVI Brazilian E-Science Workshop (BRESCI 2022)

Self Cite

View full text Add to dashboard Cite

Experimental Software Engineering has straightforwardly evolved in the last decades due to the effort of the community in providing consolidated training, teaching and practice. Particularly, for controlled experiments and quasi-experiments, the software engineering community has discussed on the lack of reproducibility and the missing of experimental artifacts sharing policies, such as, dataset, baselines, metamodels, repositories, and scripts. These are, therefore, important issues that jeopardizes controlled experimentation to evolve as rigorous as in millennial sciences as Medicine and Physics. In this ongoing work, it is presented a proposal of a conceptual framework for software engineering controlled experiments and quasi-experiments based on the main principles and practices of Open Science. It is understood that Open Science is one of the pillars to the evolution of science, consequently, to software engineering. The FAIR data, metadata, repositories, curation and provenance are some of the main practices discussed in this paper. Ongoing activities are described, in terms of how they are being performed and their relationship with prospective ones.

show abstract