What makes evidence elegant, valid, useful, and convincing? This is the question the authors of this chapter have been pursuing for the past two decades. Between us, we have published over 200 papers on empirical software engineering and organized seemingly countless panel discussions, workshops, conferences, and special journal issues about the topic.
In this article, we want to reflect a little. We review the progress in software engineering so far and ask, what’s it all good for? And where do we go from here? Our findings were that:
The quest for convincing evidence is much, much harder than we first thought.
Software engineering researchers need to work together more as a community to share more of our evidence.
We need to acknowledge that evidence has a context-specific component. Evidence that is convincing to us might not be convincing to you. Accordingly, we need to be armed and ready to reinterpret and reanalyze, if ever the audience changes.