A model to transfer knowledge from research to software engineering practice based on rapid reviews and evidence briefings

Abstract

Inspired by the promising results of Evidence Based Medicine (EBM), Evidence Based Software Engineering (EBSE) was introduced more than one decade ago. One of EBSE’s main goal is to integrate the best research evidence with Software Engineering (SE) practice. However, some claim Secondary Studies (SSs) — one of the main products of EBSE — lack connection with SE practice. To better understand this claim, the first contribution of this thesis is an analysis on how SSs are related to problems faced in SE practice. We introduced a coverage technique to compare topics investigated by SSs and issues practitioners report in Stack Exchange Q&A websites. We could observe a low coverage (15.6%), which corroborates with the claims about lack of connection between SSs and SE practice. Motivated by the findings of our first contribution, and inspired by recent advances in EBM, we propose a model to transfer knowledge from SSs to SE practice. The model adhere to two of the most prominent approaches proposed in EBM: Rapid Reviews (a lightweight research method that focus on understanding practitioners problems, and mitigating them with evidence delivered in a timely manner), and Evidence Briefings (one-page documents summarizing the main findings of an empirical study). To evaluate the EBs, we created 12 of them based on one SS each, and conducted two surveys. One with the authors of the SSs, and other with practitioners that reported issues considered as covered by the SSs. Both practitioners (32) and researchers (7), positively evaluated the EBs. We also invited authors of papers published in four SE conferences to create EBs of their papers. We conducted another survey to evaluate their perceptions. The positive results show that EBs can play a role as a transfer medium transferring knowledge from research to SE practice. To evaluate the applicability of RRs in a real context, we conducted two RRs in two software companies. The problems reported by the practitioners were due to low customer collaboration (Company 1) and low team motivation (Company 2). Following RRs strategies, we searched for evidence that might help to mitigate the problems, synthesized the results, and introduced the findings to practitioners through EBs. Through a series of interviews, we evaluated the practitioners perceptions about the introduction of RRs to support decision-making. Our results show that practitioners are rather positive about RRs. In summary, and based on the positive results we obtained with the six empirical studies we conducted, we believe the model we proposed to transfer knowledge based on RRs and EBs has a promising future as an alternative to make research closer to SE practice.