What Predicts Software Developers’ Productivity?

One of the most cited research papers from the last few years, authors Murphy-Hill et al analyze which questions are most likely to distinguish self-reported high performers from the least productive developers:

Here was the full list of how each question performed, sorted by how closely they correlated with high productivity:

The three questions most correlated with productive developers:

  • Job enthusiasm (F1)
  • Peer support for new ideas (F2)
  • Useful feedback about job performance (F11)

The three questions most negatively correlated with productive developers:

  • Personnel turnover on my project is high (F48)
  • Extensive documentation is required to use my software at different points in its lifecycle (F47)
  • My software’s platform (e.g., development environment, software stack, hardware stack) changes rapidly (F46)

I thought the design of this study was extremely well-considered, even if they were using lousy metrics like Lines of Code and Commit count. This study could be fun to try to replicated with Line Impact as the productivity metric used. I would expect similar questions to get sifted out, since the questions they found correlate most strongly do intuitively feel like elements I’ve appreciated in jobs I enjoyed.