Abstract
Many results in the social sciences fail to replicate. This means that when experiments of published results are replicated with new data, original positive results, typically indicating some treatment effect, often turn out as null results. There are many potential explanations for unsuccessful replication outcomes, including the use of “researcher degrees of freedom” to find statistically significant results and low statistical power. There are also many potential remedies to the replication problem, including pre-analysis plans and publishing studies in the “Registered Report” format. In this chapter, we go through replication failures and discuss how to do replicable research.