From the Asylum to the Prison: Rethinking the Incarceration Revolution

The incarceration explosion of the late twentieth century set off a storm of longitudinal research on the relationship between rates of imprisonment and crime, unemployment, education, and other social indicators. Those studies, however, are fundamentally flawed because they fail to measure confinement properly. They rely on imprisonment data only, and ignore historical rates of mental hospitalization. With the exception of a discrete literature on the interdependence of the mental hospital and prison populations and some studies on the explanations for the prison expansion, none of the empirical work related to the incarceration explosion – or for that matter, older research on the prison-crime and prison-unemployment relationships – takes proper account of mental hospitalization rates. When the data on mental hospitalization are combined with the data on imprisonment for the period 1928 through 2000, the incarceration revolution of the late twentieth century barely reaches the level of institutionalization that the United States experienced at mid-century. The highest rate of aggregated institutionalization during the entire century occurred in 1955 when almost 640 persons per 100,000 adults over the age 15 were institutionalized in asylums, mental hospitals, and state and federal prisons. In addition, the trend line for aggregated institutionalization reflects a mirror image of the national homicide rate during the same period. Using a Prais-Winsten regression model that corrects for autocorrelation in time-series data, and holding constant three leading structural covariates of homicide, this paper finds a large, statistically significant, and robust relationship between aggregated institutionalization and homicide. These findings underscore, more than anything, how much institutionalization there was at mid-century. The implications are both practical and theoretical. As a practical matter, empirical research that uses confinement as a value of interest should use an aggregated institutionalization rate ...