White CTSafeConnect Contacts Based on Region and Population Density

by Timothy Hall and Peter Keigher

Last update: 19 January 2022

for Data Visualization for All
with Prof. Jack Dougherty
Trinity College, Hartford CT, USA


The CCADV is the Connecticut Coalition Against Domestic Violence and its main role is to serve as a voice for victims of domestic violence and those organizations that serve them. Their CTSafeConnect program uses different forms of electronic communication as a way to make it easier for victims of domestic violence to access information, resources and assistance. Over 2020 and 2021, the organization collected data on all new individuals that contacted them on their hotline which includes race, gender and region to name just a few of the columns. The data has since been cleaned and masked to protect the identity of the people involved and allow for more accurate findings. The towns in The State of Connecticut were then grouped into 14 regions.

By this time, the class had learned some of the skills necessary to tell and visualize a data story which made us want to answer how SafeConnect contacts that identify as white differ between regions with lower and higher population density. Our research question is, how do SafeConnect contacts that identify as white differ between regions with lower or higher population density? The answers that we find will give insight into where The CCADV could consider spending more time. It will also tell if there are any low density regions that have a disproportionately high number of white contacts. This will allow for CTSafeConnect to work more efficiently in the future as using this information could lower future contact numbers if they allocate the necessary resources in certain regions. It is important for us to be able to find out if some regions are disproportionately affected because the findings from this data could lead to the discovery of underlying factors that we may not yet know about which could help reduce the number of contacts. Some of our key findings were that the Middletown region has a relatively low population density but they still account for the greatest number of first instance white contacts. Our scatter chart was able to show us that CTSafeConnect receives the greatest number of first instance interactions from regions with an average level of density which is represented by gray. This information is really important for CCADV because they can interpret it and understand that it may be valuable to look into the Middletown region to see if there is something that is causing the relatively high number of first instance interactions with whites. There are also a number of regions that the data shows receives a relatively normal amount of first instance interactions which is useful for the service because it could allow them to narrow their focus. In all, there are a number of things that can be taken away from the data to improve the performance of the CCADV.


First, we needed to find, download and clean the data to eventually be able to chart Population density of regions in Connecticut map. To do so, we extracted data from The US Census 2015-19 using The Social Explorer tool online. From here, we were able to determine the land area in square miles and population of each town in Connecticut. We then added individual town’s numbers into their proper region to determine the total number of people per square mile in each region (total population/land area). Next we used the Datawrapper tool to map this data and used tips from HandsOnDataViz to meet the proper standards and principles of creating maps. This included using the tooltips function to control what is displayed in the hover menu and using a diverging color sequence. For the other half of the data we needed to look at we created a pivot table. The rows, columns and values were adjusted to display the number of contacts by race in each region. We then used the filters tool. Under ‘2YearInstance’, the status bar was adjusted to show only one item which was ‘1’ which is the computer's way of saying it will consider only the first instance of an interaction. This method was used because it more shows new individuals interacting with CCADV. This is more valuable because considering the number of repeat contacts wouldn't measure new individuals interacting with the service. At last, we used the Datawrapper tool to chart this data in a scatter chart that showed First instance contacts with whites by region.

Region Pop. per sq. mile White contacts
Torrington region 143.66 160
Killingly region 200.28 216
Mansfield region 317.33 189
New London region 402.16 159
Middletown region 442 545
Enfield region 541.72 182
Danbury region 664.43 140
Waterbury region 1,032.12 164
Ansonia region 1,054.52 203
Hartford region 1,087.10 327
Meriden region 1,391.23 508
New Britain region 1,445.94 369
New Haven region 1,566 221
Stamford region 1,810.14 163
Bridgeport region 2,301.23 236

Some of our key findings from the first half of research when we were calculating population density was that the most densely populated regions were in the southwest tip of the state that is closest to New York City. These regions were Bridgeport and Stamford. We were also able to find that there is a larger % of the state in terms of square miles that is considered low density. When comparing what each map tells us is when you learn the most in this case. For example, the map is telling us that the Middletown region has a relatively low population density when compared to other regions. Then, if you take a look at the chart, you are able to see that the Middletown region has low density but accounts for the greatest number of first instance white contacts.

One of the things that we were able to determine from the scatter chart is that regions that were classified as having a mid level of density had the most number of first instance interactions with those identifying as white on average. This idea is especially true in the case of Meriden, New Britain and Hartford. We initially started out defining population density by two terms: urban and rural but found that these terms did not accurately describe the regions. For example, Hartford was considered rural in our first calculation which is inaccurate. Thus, it was more appropriate to define it in terms of level of density.

When understanding our data and findings, it is essential to keep in mind the context of the data story. As described, the high and low density regions were divided up based on how the data was organized with our partners, making some regions appear to be more or less densely populated than some might expect. We found the Hartford region, for example, has a mid level population density even though it has one of the most populated cities. This is just how the state is divided up specifically to this data story. It is also important to note that some callers did not identify their race, so although we have everyone that did identify as white, it is likely that there are more white callers that have not identified their race. We do not think this skewed any of our data and findings drastically.


Thank you CTSafeConnect, specifically Maria Guzman and Joanne Vitarelli, for the Data and guidance throughout the process. Pleasue use this contact information if you or a loved one is suffering from domestic violence.

Phone Call: 1-888-774-2900 Email: safeconnect@ctcadv.org Text: 1-888-774-2900 Live Chat: Link available at https://ctsafeconnect.com