Ulysses' Pact or Ulysses' Raft: Using Pre-Analysis Plans in Experimental and Non-Experimental Research

In recent years, pre-analysis plans have been adopted by economists in response to concerns raised about robustness and transparency in social science research. By pre-specifying an analysis plan, researchers bind themselves and thus avoid the temptation to data mine or $p$-hack. The application of pre-analysis plans has been most widely used for randomized evaluations, particularly in the field of development economics. The increased use of pre-analysis plans has raised competing concerns that detailed plans are overly restrictive and limit the type of inspiration that only comes from exploring the data. This paper considers these competing views of pre-analysis plans, examines the extent that pre-analysis plans have been used in research conducted by agricultural economists, and discusses the usefulness of pre-analysis plans for non-experimental economic research.