In an earlier post, we showed how rural prisons and census methods create distortions in county demographic and health statistics. In this post, we add further evidence and discuss the economic, political, and research implications of these potentially misleading data.
National survey methodology may contribute to inaccurate data
Ohio State Reformatory, Mansfield
Census policy coupled with national survey methodology may skew the reporting of rural data and their effective use. For example, state and local programs often use data from the Behavioral Health Risk Surveillance System (BRFSS) to evaluate outcomes, target health promotion and disease programs, and monitor changes. Most states use the BRFSS to inform legislative priorities. However, BRFSS does not survey, and thus cannot account for, incarcerated populations.
Texas and Colorado as examples
For example, two counties in Texas stand out due to their 15-25% lower birth rates than the surrounding counties – because both are home to large prisons for women.
Also, BRFSS estimates the rate of heavy drinking in Colorado as being 6.5%, with Logan County listed at 5.8%. This rate does not account for Logan’s 2,585 inmates. Removing prison inmates from the denominator results in an estimate more in-line with the state average. Note: 5.8% of 22,000 = 1,276 heavy drinkers; 22,000 – 2,585 inmates = 19,415; hence 1,276/19,415 = 6.6%.
Finally, Bent, County Colorado was reported to have the third-highest increase in smoking rates between 1996 and 2012. This county of 6,000 residents had a Veterans psychiatric unit which converted to a prison for 500 inmates in 2001. The facility then transitioned to a state rehab facility for homeless individuals suffering substance use disorders.
The authors reported that “These local, annual measurements of cigarette smoking prevalence can be an important stimulus to local public health decision-making and community engagement.” Bent County is small and likely has statistically unstable BRFSS sampling data. Without considering these institutionalized populations, it is impossible to obtain true tobacco use prevalence in this rural county. Yet, the label of “third-highest” and associated fallout for the county remain. Analysis of BRFSS data without consideration of prison populations will lead to inaccurate reporting.
Economic, tax, and legislative districting implications
In the 1970s, it was believed that prisons might provide economic relief from agricultural, mining, and out-migration losses. The limited economic development from prisons may contribute to a negative association with being a “prison town” or offset by misinformed state and national health policy. At best, national analyses of the potential benefits from a prison in a rural county [PDF] show mixed results.
A significant portion (27%) of revenue from Colorado’s tobacco tax is distributed to local governments based on the amount of revenue collected within a given city or county. For counties with higher levels of the incarcerated, their tobacco tax revenue will seem lower than expected, leading to the potential for decreased funding distribution by the state.
Also, the impact of counting prison inmates in their county of incarceration relates to local and state legislative districting. In counties with large prison populations, relatively small numbers of voting citizens carry disproportionate voting power, leading to significant political conversations.
Rural prisons are just common enough to confound rural health services research
Approximately 350 US rural counties have a prison that accounts for 5-25% of their populations. Large incarcerated populations in rural communities affect county demographics, reported health behaviors, and other statistics. Counties with large prisons stand out on maps because of the contrast with neighboring counties. This can lead to erroneous interpretation of regional variations in demographics and healthcare, leading to misinformed policy recommendations. As noted above, such errors and misclassification can affect state and federal funding allocations, and stigmatize communities. Attributing inmates to the census tract in which they lived before incarceration — or where they were sentenced — could correct the skewing of demographic and health-related statistics in these rural communities with large prisons.
Rural research requires local knowledge
It is easy to fold the sparse communities of rural America into larger populations and datasets. However, doing so risks losing track of local characteristics. This can have profound impacts on resultant analysis and interpretation. Just as prison populations can affect rural community health data, other entities such as colleges, large single employers (such as meatpacking plants), and geographic isolation may also have influence. The term “rural” is about more than just proper numerators and denominators in rural health services research and policy making. Researchers should consider the impact of prison populations and account for local nuance in their data and their work.
Jack Westfall
Latest posts by Jack Westfall (see all)
Hoon Byun