Reading focus
Syllabus: How Can Psychological Science Become More Reliable?
Trace the model’s selection mechanism and assess its scope alongside the linked correction.
Why you’re reading it. Consider why individual intentions alone may not explain patterns in a scientific record.
A question to consider. Which institutional practice would you change to test the proposed mechanism, and what outcome would count against it?
Why this reading
Research practices can spread through institutional selection without anyone intending to produce unreliable findings. Smaldino and McElreath model how rewards for publication quantity could change the methods that persist across generations of laboratories.
Before you read
An agent-based model represents interacting units with explicit rules. Here the units are laboratories. Modeled “power,” effort and success simplify a research process; they are not direct measurements of every real laboratory.
Read for
- Connect assumptions about effort, output and inheritance to the simulated patterns.
- Distinguish a demonstration of a possible mechanism from evidence about its prevalence.
- Consider which omitted incentives or alternative rules could change the conclusion.
Consider afterwards
- What observations would help determine whether the modeled selection process operates in a particular research institution?
Editorial context
A 2023 correction addresses a coding error affecting replication-rate dynamics, a sequencing discrepancy and typographical errors. Kohrt and colleagues reimplemented the model: some trajectories changed while the broad conclusion persisted. The simulation is not a field experiment on incentives.