Abstract
Lab and online experiments are widely used tools in economics, as well as in other areas of the social sciences. They are typically designed either to test treatment effects or to elicit parameter values for concepts such as economic preferences. While the role of pre-registration of designs and analyses is increasingly discussed in workflows, many other aspects of the research process are less visible in the communicated output. We outline what we view as the common workflow for lab and online experiments in economics, highlight the steps that we believe are often 'missing' and discuss how these omissions may undermine the replicability, credibility and generalizability of published findings. This article is part of the theme issue 'Statistical workflow'.