These experiments have been popularly cited for the propositions that arrest does not deter abusers who are either employed or unemployed. This analysis reviewed those experiments and found that they were fundamentally flawed by failing to consider that domestic violence, if left unchecked, usually escalates in severity and frequency and by failing to account for the impact of decisions by prosecutors and courts subsequent to arrest. Additional design flaws, in which investigators included cases of self-defense and assigned bad-risk perpetrators to the arrest category, further undercut the validity of the results. This analysis notes that other published studies show that arrest is generally the superior method of deterring future violence. Furthermore, these police replication studies did not recognize that arrest in domestic violence situations allows the victim a window of opportunity to secure safety and to contact appropriate services. 278 notes and 1 appendix