The neuropeptide oxytocin has garnered considerable interest for its role in social behavior and its potential for the treatment of psychiatric illnesses characterised by social dysfunction, such as autism. However, initial excitement has turned to disappointment with some studies failing to replicate earlier results, which has been attributed to issues surrounding research methods, mechanistic understanding, and theory development. In this talk, I discuss efforts to improve research methods to enhance reproducibility, including precise sample size estimation, synthetic datasets, and ways to test evidence for null models