Which practice most directly improves the credibility of experimental causal claims?

Pre-registering hypotheses and methods before data collection

Selecting archival datasets based solely on availability

Using only open-ended questions to capture nuance

Reporting only significant findings to emphasize strong effects