Table 1. Parameters used in the simulations.
Values in italics are the baseline scenario.
| Parameter | Definition | Values tested |
|---|---|---|
| Parameters from Smaldino and McElreath (2016): | ||
| η | Strength of the association between effort (e) and the probability of tackling a new hypothesis | 0.2 |
| Wi | Power of a lab to correctly detect a true hypothesis | 0.8 |
| e0 | Starting effort for labs | 10, 50, 75 |
| ψ | Random mutation in effort from a standard Normal distribution | N(0,1) |
| Additional parameters for simulations presented here: | ||
| j | Cycles per audit | 50, 75, 150, 200 |
| n | Number of papers needed to be eligible for auditing | 25, 50 |
| el+ | Increased effort of audited labs | 0, 1, 5, 10 |
| en+ | Increased effort of networked labs | 0, 0.5, 1, 5, 10 |
| c | Cut-off in cumulative distribution function at which labs are removed | 0.67, 0.95, 1 |
| Review error | Percent by which the number of false positives are randomly changed in order to simulate reviewer error | 0, 20 |
| Mistake probability | Mistake probability for papers with a false positive finding | 0.25 |