Politics and historical imprisonment, 1930–2010

Abstract

Theorists have argued that the increasing reliance on imprisonment in the late twentieth and the early twenty-first centuries is associated with a number of important social and political developments that began in the 1960s and 1970s. Despite the emphasis that has been placed on these historically contingent developments, scholars have yet to empirically examine whether the predictive power of recent political variables is restricted to the use of imprisonment in the late twentieth century. The current study addresses this gap in the literature by examining whether four political perspectives are able to account for the use of incarceration at the state level from 1930 to 2010. This study also examines whether the movement from the early to the late twentieth century changed the nature of the relationship between political considerations and jurisdictional reliance on imprisonment. The results indicate that the political party affiliation of elected officials, the public’s commitment to religious fundamentalist sentiment, and the size of African-American populations are significant predictors of imprisonment across the twentieth and early twenty-first centuries.

Keywords:

• Incarceration
• penal policy
• crime politics

Notes

1. Nebraska has embraced a unicameral, non-partisan legislative system from 1937 to the present, and Minnesota used non-partisan ballots from 1913 to 1974.

2. The incarceration rate for Delaware was not reported for the measurement period in 1932. The incarceration rates for 1931 and 1933 were averaged to provide an estimate for this missing year.

3. The religious affiliations of citizens were not reported in the 1960s. To avoid excluding this decade from the analyses, estimates for this measurement period were calculated using the average percentage of Protestant fundamentalists from 1952 and 1971. In addition, the collection of jurisdictional religious affiliations did not occur during decennial measurement periods prior to 1980. The closest collection periods are used for each decennial measurement point for the decades from 1930 to 1950. Research by Morgan and Watson (1981) indicates that this measurement strategy is appropriate because jurisdictional religious affiliations are relatively stable over time.

4. Violent and property crime rates for the measurement period in 1930 were drawn from the UCR report for 1931. Due to the initiation of the program in 1930, there is a substantial amount of missing state-level data for this year. Furthermore, jurisdictional data on rape were not reported in the 1940 UCR report; therefore, the violent crime rate for this measurement period consists of the rate of homicide, robbery, and aggravated assault per 100,000 individuals.

5. To determine the variable that increased the predictive value of the percentage of Republicans in the state legislator indicator, this measure was included in its respective individual model; the remaining theoretical predictors were then added one at a time. The results from this analysis (not shown) indicated that the predictive value of the percentage of Republicans in the state legislature measure increased with the inclusion of the religious fundamentalist variable. Additional analyses were conducted to determine the nature of the relationship between these variables. In these analyses, an interaction term was created using the percentage of Republicans in the state legislature and the percentage of fundamentalists in the population variables, and this product term was then introduced into the fully specified model. The results from this analysis (not shown) indicated that the interaction term is significantly and positively related to jurisdictional imprisonment (p < 0.001). This finding indicates that the percentage of Republicans in the state legislature is more strongly related to imprisonment in jurisdictions with a higher percentage of religious fundamentalists in the state population.

6. One limitation associated with the use of historical data involves reliance on early twentieth century UCR crime rates. As numerous scholars have noted, UCR crime rates before 1960 were influenced by non-reporting, under-reporting, the adoption of different statutory definitions across jurisdictions, and the use of decennial population counts when calculating yearly crime rates between the decennial measurement periods (Mosher, Miethe, and Hart 2011; Skogan 1975; Wolfgang 1963). Based on these limitations, the results for the violent and property crime rate variables in the fully specified model and the split-half model for 1930 to 1960 must be taken with some caution.