Solitary confinement is often used as a form of punishment for inmate misconduct, a practice known as disciplinary custody. One justification for the use of disciplinary custody is that it should deter future misconduct by increasing the perceived costs associated with committing an infraction. However, research on the deterrent effect of disciplinary custody is limited, largely due to a conceptualization of it as a singular experience, which ignores significant heterogeneity within that punishment. The few studies that have examined one type of treatment heterogeneity, length of stay, are limited in the grouping of varying forms of isolation, such as administrative custody and disciplinary custody, or in the scope of behavior examined post-release from disciplinary custody.This dissertation built on past studies by examining two types of treatment heterogeneity: (1) length of stay by focusing specifically on disciplinary custody and expanding on the types of misconduct (beyond violent acts) considered post-release from disciplinary custody; and (2) an early release mechanism. With data from a large state correctional system, this study utilized a sample of first time admissions from 2012 to 2014 who experienced a disciplinary custody stay, and their institutional outcomes were followed through August 2017. This study used inverse probability weighting with regression adjustment, including a large array of relevant covariates to account for pre-existing differences in the treatment conditions examined. The results of this study do not support specific deterrence theory justifications for the use of disciplinary custody. There was no evidence that increased severity of disciplinary custody stays, either through longer lengths of stay or through serving more than the original sanction length assigned, resulted in lower likelihoods of subsequent misconduct or fewer days until a subsequent misconduct. Implications and future directions are discussed.