Box 1.
10 Reasons to replicate simulation studies.
| Reasons to replicate simulation studies | Explanation |
|---|---|
| 1) Major impact | Highly cited simulation studies can influence many subsequent studies and form the foundation of data analysis across different research fields. |
| 2) Conflicts of interest | Researchers conducting simulation studies may be invested in certain methods which may bias design choices and how result are presented. |
| 3) Selective reporting of results | Journal restriction may limit the amount the result being presented, yet favoritism toward one method can bias focus of reported results. |
| 4) Competing aims | Results of simulation studies are relevant for different audiences: while methodologically oriented readers may be interested in general properties of investigated methods, applied researchers may look for guidance for their particular use case. |
| 5) Coding errors | Although coding errors can happen to anyone, there is generally still a lack of code review and often unavailability of simulation code that allows for (external) checking code. |
| 6) Limited scope | Since the number of simulation scenarios is finite, generalizing to a particular research setting be beyond the scope of the simulation study, thus requiring replication for these further scenarios. |
| 7) Importance of details | Reported information may be insufficient for comprehensive assessment of (results of) simulation studies - even for dedicated peer-reviewers. |
| 8) Insights as individuals and as a field | Replication encourages reflection on reporting standards and practices such as making code publicly available, code review, and pre-registration of simulation studies. |
| 9) Lead by example | Methodologists have the chance to practice what they preach. |
| 10) Because we can | There are no financial, logistic, historical, or ethical constraints to replication (although there are often serious time and funding constraints to individuals that produce them!) |