Assessing attitudes towards evidence-based software engineering in a government agency

Context: Evidence-based practice (EBP) has allowed several disciplines to become more mature by emphasizing the use of evidence from well-designed and well-conducted research in decision-making. Its application in SE, Evidence-based software engineering (EBSE) can help to bridge the gap between academia and industry by bringing together academic rigor and research of practical relevance. To achieve this, it seems necessary to improve its adoption.Objective: We sought both to study the attitudes towards EBSE of stakeholders working in a government agency (GA) and to assess whether knowledge of EBSE would impact their working practices. Method: We conducted a multi-stage field investigation in an Uruguayan national GA that is responsible for digital policies. First, we organized an EBSE awareness lecture and we collected and analyzed participants' perceptions of the value and limitations of EBSE. Sixteen months later, in a second stage, we contacted the agency and asked participants whether they had made use of the information about EBSE we presented to them.Results: Initially, participants reported that EBSE seemed useful for tackling challenging problems and, in particular, considered its use appropriate given the agency's responsibilities. Perceived barriers to EBSE adoption were the need for institutional support, the lack of government practice reports, inadequate skills or motivation, the cost of conducting systematic reviews, and the lack of evidence about emerging issues. In the follow-up survey, although the participants were not undertaking systematic reviews themselves, many reported improvements in how they searched for and evaluated information to support their work.Conclusion: Our study presents some insights to better understand EBSE adoption. With the exception of GA-specific issues, perceived value and barriers to adoption were consistent with those reported in software engineering and other disciplines. Our follow-up study confirms the potential value of evidence in the context of IT regulatory and government bodies.