Chronic Crime Behavior through the lens of Mental Health and Other Social Dimensions:
An examination of the 2016 Inmates Survey from the Bureau of Justice Statistics
By Jesica Ramirez
Mass incarceration represents a public health and social issue in the U.S, particularly, the high levels of recidivism are concerning. According to the National Institute of Justice, more than 60% of the prisoners released in 2005 were arrested within 3 years and 83% within 9 years.

In an effort to further study and address this problem, we examine the rates of recidivism across several social dimensions such as mental illness prevalence, emotional difficulties and family conditions (e.g. foster care, homelessness), using the data from the BJS 2016 Inmates Survey.

Inmates with Any Mental Illness show higher number of arrests.

The following graph shows the average arrested times across different age groups. In general, younger groups have higher rates of arrests, which is no surprise (See Cook and Laub 1998). The main point of this graph is how the curves corresponding to inmates with serious and non-serious mental illness are evidently higher of those inmates who reported no mental illness. And the differences are wider if we only look at the female population with any serious mental illness. (Click on the button below)
Select Gender Population
This is relevant because less than half of the inmates who report a mental illness, have received any treatment in jail or prison, and the rate is lower for inmates that continue to receive treatment (BJS 2016 Inmates Survey), which is concerning as all prisoners have a constitutional right to health care. Additionally, the high levels of arrests for inmates with a mental illness highlight the need to improve the access to mental health services as it might prevent further arrests and incarceration. (See Wilper et al. 2009)

Analyzing other mental health dimensions and family conditions.

Important to note that in the survey, inmates voluntarily shared information about their mental health and diagnosed with mental illness by an MD. People that live in high-risk conditions, like most of the prison population, are also less likely to have access to medical resources (i.e. medical appointments, health care treatments), and so the likelihood of being diagnosed by an MD is lower as the access to health care is lower. This implication is relevant because in the data we only see voluntary reported and diagnosed mental illnesses. Hence, another variable to look at is emotional distress and how this affects recidivism. Also, criminological theories have long suggested that family relationships influence criminal behavior. In this sense, we study several family conditions of the inmates like being in foster care, having a family member in prison, being homeless in the past, and being homeless before turning 18.




On average, the higher frequency of restlessness on inmates, the higher number of reported arrests.

In the survey, inmates were asked how often have they feel restless or fidgety during the past month The responses vary from "None of the time" to "All the time". In this graph, we show the average levels of arrests across the reported frequency of restlessness by age groups.
In general, the color (number of arrests) of each column representing an age group gets darker as we move from a low frequency of restlessness to a high frequency.
This is particularly true for young to middle age groups. For example, looking at the group of 35-44 years old, inmates who reported a zero frequency of restlessness ("None of the time") had on average 8.02 arrests, whereas inmates that reported a high frequency ('All the time') of restlessness had on average 12.20 arrests.
With this graph, we show again how mental health, studied as emotional difficulties, also shows a correlation with high levels of arrests.


Hispanic and unspecified race groups show larger disparities in the number of arrests across reported emotional difficulties.

Similar to the graph above, this one is measuring average arrests across the frequency of restlessness. However, in this graph, we look across different race groups. The general message remains the same, as we move upwards (higher frequency of restlessness) for each race group, the levels of average arrests increase.
However, across race, we note that Hispanic groups show higher disparities: hispanics who reported no restless feelings, had on average, 6.17 arrests, whereas hispanics with the highest frequency of restlessness had 12.55 arrests, two times more.
Groups of inmates who had an unspecified race also reported high disparities of arrests across the restlessness levels.
In the survey, inmates were also asked about their frequency of hopelessness and nervous feelings. Although we don't report it here, the pattern stays the same, higher levels of emotional difficulties are related to higher levels of arrests across age and race groups.

What about past family conditions?

Note: For this graph, categories in the y-axis are different and do not represent ordinal values like the previous heatmap charts.

Some criminological theories suggested that family relationships influence criminal behavior (See: Akers & Sellers 2013). Given that the survey collected some information about social and family characteristics of the inmates, we explore the levels of arrests across five different family situations or conditions: being homeless before the age of 18, being homeless in the past 30 days before the arrest, being in foster care and having any direct family members in prison which includes:partner, mother, father, children, or siblings.
Note that these categories are not mutually exclusive, an inmate could have been in foster care and also have any direct relative in prison. We also include a category called "Regular Average", which refers to the overall average (bottom row) of arrests across all inmates (with or without any particular family condition).
Each square represents the average level of arrests for a given age group and family condition. For example, the upper-left square shows an average of 8.82 arrests for the young group of 18-24 that have been homeless before 18. As we move in this column, for this group of 18-24-year-olds, the ones that have being in foster care show the highest average arrest rates in 18-24-year-olds.
An interesting result is in the 55-64 group age, the inmates who were homeless before 18 had on average 19.9 arrests, two times the overall average of that group of 8 arrests.

Being homeless: a relevant factor on recidivism!

Looking across race/ethnicity groups, for Hispanic, white, and other races, inmates that have been homeless before 18 show the highest average arrests in each group. For black inmates, those who experienced homelessness at least 30 days before the arrest show the highest average arrests among black.
This reinforces the fact that homeless people are over-criminalized. Particularly, black people are at far greater risk of being targeted, profiled, and arrested for minor offenses. Additionally, along with systemic housing discrimination, black people exiting jails and prisons often face significant problems in accessing safe and affordable housing which leads to a vicious circle of being homeless and potentially arrested again.

Do inmates have access to mental health support? Why is this relevant?

According to the 2016 Inmates survey, less than half of the inmates with a diagnosed mental illness receive mental health support at prisons or jails. Here we show how people with mental illness reported, on average, higher levels of arrests. If we look at emotional distress variables (i.e. restlessness), again inmates across age groups (particularly young to middle age inmates) who reported higher levels of emotional distress, also report on average higher arrests. And a similar story is explained by inmates with difficult family conditions, especially people who experienced homelessness before the age of 18.

All prisoners, including those with mental illness, have the right to be treated humanely and with respect for their inherent dignity as human beings. According to the World Health Organization, "addressing mental health needs will improve the health and quality of life of both prisoners with mental disorders and of the prison population as a whole. By promoting a greater understanding of the problems faced by those with mental disorders, stigma and discrimination can be reduced. Ultimately, addressing the needs of people with mental disorders improves the probability that upon leaving prison they will be able to adjust to community life, which may, in turn, reduce the likelihood that they will return to prison".

As Bryan Stevenson states, we must embrace "the challenge of creating a fairer, more merciful criminal justice system". Taking this approach would not only give opportunities to inmates to change their lives but also lead society to inclusiveness and progress.

Sources

  • Survey of Prison Inmates (2016) from the Bureau of Justice Statistics
    Notes: For this data storytelling, we use the final analysis weight variable (V1585 - WT_FINAL) of the survey to calculate the average number of arrests in each group combination.
    To obtain the aggregated data just click here. The original survey is on the link above.
  • Akers,Ronald L. & Sellers, Christine S. (2013) Criminological Theories: Introduction, Evaluation, and Application. Oxford University Press
  • Cook, Philip J. & Laub, John H.(1998) Crime and Justice, Vol. 24, Youth Violence. The University of Chicago Press
  • Mental Health and Prisons. World Health Report 2001: Mental Health: New Understanding, New Hope. Geneva, World Health Organization.
  • Pettus-Davis, Carrie. & Epperson, Matthew W. (2014) From Mass Incarceration to Smart Decarceration. Center for Social Development.
  • Thompson, C. & Eldridge, T. (2018) Treatment Denied: The Mental Health Crisis in Federal Prisons. The Marshall Project.
  • Wilper AP, Woolhandler S, Boyd JW, Lasser KE, McCormick D, Bor DH, Himmelstein DU. (2009) The health and health care of US prisoners: results of a nationwide survey. Am J Public Health.
  • Code References:
    Observable and dear old friend: Stack Overflow.
    Scrollama for the scrollytelling
    Simple tooltips in v6 , Heatmap with tooltip in d3.js , d3 mouseover multi-line chart

Acknowledgements