Crime and community heterogeneity: race, ethnicity, and religion

Abstract

Prior studies have explored the relationship between race and religion and their effect on various crimes. In the USA race is typically defined as the proportion of a community that is African-American or nonwhite. Likewise, religion is defined as the proportion of a community that adheres to any religious denomination. This study extends earlier work by employing Herfindahl indices as measures of community homogeneity with respect to race and ethnicity as well as religious denominations. It also measures religiosity based on four different denominational groups, rather than religiosity as an aggregate. Results indicate that as a community's degree of homogeneity increases, in terms of both race/ethnicity and religion, crime decreases; and that the effects of religion on crime may vary by denomination.

Notes

¹ Despite our sound theoretical reasoning, replacing percent black with our race and ethnicity Herfindahl results in only minor improvements in goodness of fit for the models described. The lack of significant improved fit is an artifact of the racial and ethnic homogeneity of Kentucky. Percentage black and the Herfindahl have a correlation of −0.97. Using state-level race and ethnicity data, the correlation between percentage black and the Herfindahl measure is −0.44, suggesting that there may be a larger impact on estimation results when replacing percentage black with our Herfindahl measure in other states. Furthermore, Kentucky stands out as a fairly homogeneous state. Kentucky, with a Herfindahl of 0.80 ranks 8th in the nation. The top five states are Maine, Vermont, New Hampshire, West Virginia and Iowa with Herfindhal values ranging from 0.93 to 0.86. The most diverse are Hawaii, California, New Mexico, Texas and New York with values ranging from 0.23 to 0.42. Future research would lead to testing the measure of racial and ethnic homogeneity versus percentage black or percentage minority in more heterogeneous state(s).

² We predicted that all denominations would exhibit a negative impact on crime rates. Surprisingly, the share of evangelical Protestants, the primary denomination in Kentucky, does not have a statistically significant effect on any of the crime rates. Further analysis suggests that this result is primarily due to the interaction of evangelical share with the religious Herfindahl and population density. The sign of the coefficient on cep becomes negative when omitting the religious Herfindahl. Note that evangelical Protestants is the primary denominational group in most counties. The additional omission of population density results in a statistically significant coefficient on cep. This anomaly does not affect the other religions. It appears multicollinearity for these three variables results in insignificant results on cep.