Wiemers, Elizabeth AnneThis paper examines the utility of using a unique variation produced by a ruling in the Maryland sentencing guidelines as an instrumental variable in future research. The guidelines specify that at age 26 an offender's juvenile record is discounted from the calculations, resulting in a lower sentencing grid placement and a shorter sentence. I examine in depth the appropriateness of this treatment rule as an instrumental variable for research and find that it is an effective instrument. The study also examines preliminary results produced by using an instrumental variable to estimate a relationship between sentence severity and recidivism. The use of instrumental variables corrects for the selection bias present in other studies in this area by allowing me to compare individuals affected by the treatment rule. The results indicate a slight deterrent effect of imprisonment; overall, increasing sentence length decreases future rates of recidivism.en-USThesisSociology, Criminology and Penology