Introduction

Inspired by this 2016 Marshall Project article and Youth First Initiative report about the eighty remaining youth prisons in the United States, I wanted to look more closely at the geographic connection between schools, incarceration, and county population demographics in my home state of Arizona. There are many more incarceration facilities for youth in the United States; the eighty featured are at least one hundred years old and fit more of a traditional prison structure. The non-profit and non-partisan Prison Policy Initiative’s report on Arizona demonstrated an increase in both prison and jail populations in Arizona from 1978 to 2015, driven largely by increases in pre-trial incarceration. This is against a more recent national trend. However, Arizona’s prisons reflect the same racial imbalance describe by the Youth First Initiative, with overrepresentation of black, Native, and latino populations. I wanted to place Arizona prisons, schools, and other insitutions on an interactive map that would allow a user to both see broader institution clusters, zoom in on particular regions, and get details about each institution.

 

Inspiration

The Youth First Initiative located the eighty oldest youth prisons in the United States on an interactive map that offered greater detail about each prison population and human rights violations when a marker is clicked.

The Arizona youth prison looked like this in the closer inspection:

I liked the use of a broad overview with specific details and the use of google street view and descriptive data that highlighted the serious impact of youth prisons.

The Washington Post published an article showing the number of incarcerated people per US county. Simple and clear the map allowed the author to discuss differences in prisoner placement between states by breaking the data down by county. The map would have benefitted from interactivity, at least the ability to zoom, to deal with the typical differences in scale between US coasts.

The University of California put California schools and adult and juvenile prisons on a map together and made a dashboard showing prison and school populations in the state.

This is the kind of dashboard that I think I would work best with the data I found; however, I wanted to work with the multiple layers in CartoDB for this project. I liked the clear colors and layout of this visualization.

 

Materials

For this visualization I used 2016 shapefiles of Arizona prisons from AZGEO Open Data and Homeland Infrastructure Foundation-Level Data (HIFLD). I also used 2016 public school, university, daycare, places of worship, mobile home parks, and day care center locations for the whole country from the same sources. I struggled to decide between showing county or census-tract level population data; as a consequence of searching for both and shapefiles for the two I ran out of time to work with broader population data. However, I used 2012 shapefiles of Arizona counties from the US Census bureau to give a better regional sense of the map.

 

Methods

I imported the data sets directly into CartoDB. Because they were already fairly clean, I did not need to do any cleaning of the data ahead of time. Once in CartoDB I assigned formats to the different data elements like date, number, or text. I decided to remove mobile home parks and places of worship from the map to unclutter the results.

I then colored and added labels to all of the remaining elements. I chose red for the prisons and labelled them with population numbers, security level, name, and capacity. It is possible that the color red has too many negative connotations and could affect viewers’ attitudes, but the color helped to locate the shapes on the map, especially against the much more prominent points assigned by Carto to the other data. I chose light green for public schools, dark green for universities, and yellow for daycare centers.

I then added name labels for all of the prisons to counteract their invisibility at a higher zoom and added a widget to filter on name of the prison.

 

Results

The resulting map had promise but also many flaws. It proved to be more of a sandbox experiment with different ways of comparing many different institutions and making their geographical relationship evident.

You can see the full interactive map here.

The broader view does not clearly show where the prisons are located, but does show clusters of schools especially. At this zoom level the number of public schools overcrowds the other institutions and the clustering seems to be based on population centers. This is where a choropleth layer of population by census region would have been helpful to contrast institutional clusters with populations.

At a closer zoom and with the pop up labels, you can more clearly see some of the ways that prisons in the small town of Florence (a high prison-density area between and to the East of the two largest cities Tucson and  Phoenix) have encouraged the clustering of day care centers public schools. The latter implies the existence of youth prison populations, which is then made evident with the label feature.

In the more populous Phoenix metro area, the map shows how local jails are slightly removed from schools and daycares, except for the specific institutions that serve their youth inmates.

While the school serving the juvenile detention center has an enrollment of 82, the dention center has a capacity of over 200. Unfortunately, the actual capacity data was missing from this data set, making the actual population versus enrollment impossible to determine in this case.

 

Future Directions

In future I think that there is a lot that could be done with this and more data. I think that a dashboard like the University of California one would work well for this Arizona data. However, I would like to have even more of an emphasis on the map, showing more detailed depictions of population numbers surrounding prisons through a choropleth map layer. I would also indicate prisons with single points instead of using shapefiles for visibility, or see if I could use both.

In future I would also try to select colors that are more appropriate for users with color blindness.