Mapping Homelessness and Gender in 2020
By David Jansen, Via Romano, and Laura Schroeder
For our final project, we aim to examine and map the homelessness problem in the United States and Territories in 2020. Our goal is to bring awareness to the housing crisis by mapping the number of homeless people, examining the demographics of the homeless population across the country, and looking at the homeless population per capita within each state. We feel this project is especially relevant given how the pandemic has affected families and housing costs.
The data in our dataset was gathered by the United States Department of Housing and Urban Development (HUD), which is responsible for addressing the nation’s housing needs and upholding fair housing laws. The HUD collected this data with a Point in Time (PIT) count, conducted on one night in January and a Housing Inventory Count (HIC), which is a point-in-time inventory of programs that provide housing, beds, and units specifically made to help people suffering from homelessness. The HUD performs these counts annually, going back as far as 2007 in our state population dataset.
However, this data is far from perfect. First, the data collected from the United States territories is also very lacking. We had to exclude American Samoa from our analysis because there was no data collected on it at all. We decided to drop it from the set as a result, but it is important to note that this data does not present a full and complete picture of homelessness across the United States and its territories as the HUD did not collect data on one of them.
Another potential source of error in our data is the fact that there is no way of knowing whether or not individuals reported their gender or are presenting their gender as they actually identify. For example, someone who identifies as gender non-conforming might instead report the gender which they were assigned at birth if they are or have been in a situation in which it is unsafe for them to reveal their identity. Further, having a category simply called “transgender” is interesting. Many people who transition specifically from male to female or female to male may be more inclined to label their gender as simply male or female, rather than the umbrella term of “transgender.” Considering that homelessness is a significant problem in LGBTQ+ youths, it is important to remember that such data that is often self-reported in nature may not paint the full picture of LGBTQ+ homelessness.
We also feel that this potential error source could be an ethical concern as well. LGBTQ+ people may have been unable to disclose their identity due to safety-related concerns, and if the HUD was not mindful of this when they collected the data, that could create a strong ethical concern. We also believe that it is important to be generally sensitive while collecting this kind of data because each number in the data the HUD has compiled represents a real homeless person in America. In addition to hoping the HUD was respectful of this fact when collecting this data, we wanted to be empathetic to and mindful of this fact when conducting our own analysis.
When analyzing our data, we used both Pandas and folium mapping. We decided to use mapping as one chosen method of analysis because we wanted to get a visual representation of the data as we felt it would effectively highlight the scale of the homelessness problem and show where it is most rampant. We also chose to use Pandas to take a closer look at the actual numbers and have a strong quantitative component to go along with our visual analysis.
First, we examined the total homeless population across states and territories, generating the map below:
The map has been cropped for size here, as zooming out to fit all the territories in the picture made the map unreadable, and therefore not helpful. Additionally, the information for the U.S. Territories seemed unreliable as the study completely left out data for American Samoa. Therefore, while this project does include data from the other territories, the map view is focusing largely on the contiguous 48 states, Puerto Rico, and the Virgin Islands.
From this map, we drew the obvious conclusion that the biggest states have the biggest homeless populations. When we used Pandas to sort the overall homeless population by state and territory in descending order, the top four states with the largest overall homeless population were California, New York, Florida, and Texas in that order. These are also the four biggest states in the country by population. However, California’s homelessness problem is disproportionately large, as it has 161,548 homeless, which is 27.83% of the country’s entire homeless population of 580,466. Washington and Massachusetts also have a large overall homeless population despite the fact that their state populations are much smaller than those of California, New York, Florida, and Texas. They were fifth and sixth in overall homelessness.
Next, we calculated each state’s homeless population per capita by dividing the overall homeless population by the state’s total population. The map of homelessness per capita looks like this:
For this map, we chose to include images of Guam, Hawaii, and the Northern Mariana Islands as all three places had high overall rates of homelessness per capita, which were meaningful to our analysis. So even though we couldn’t get a single image of the map with all the territories and states, we wanted to include them visually. Per capita, the places with the most homelessness are the Northern Mariana Islands, Washington DC, and Guam. 2.34% of the total population of the Northern Mariana Islands is homeless, while .925% of Washington DC’s and .469% of Guam’s overall populations are homeless. Islands generally seem to have high rates of homelessness per capita, with the Northern Mariana Islands first overall, Guam third, Hawaii fifth with a homelessness per capita rate of .444%, and the U.S. Virgin Islands ninth with a rate of .286%. Puerto Rico is the only island that doesn’t crack the top ten and is actually low on the list. It comes in at fiftieth overall with a homelessness per capita rate of .075%.
We then decided to take a closer look at how homelessness rates varied by gender. First of all, we were curious as to how rates would vary between men and women, and calculated the totals in addition to each map. Once that was determined, which societal factors were at play to bring us to that result? Further, we thought it was interesting how they chose to categorize the genders. The data has four categories: male, female, gender non-conforming, and transgender. As we mentioned above, this could lead to some ambiguity, as well as some potential for error, so we wanted to see how that actually played out in the data reported.
Below are the maps examining gender and homelessness:
Female Homelessness Map
Male Homelessness Map
Transgender Homelessness Map
Gender Non-Conforming Homelessness Map
After taking a closer look at these maps together, there were many similarities and differences between them. The male population of homelessness was, in total, greater than the female number. There was a large difference in Florida, specifically with 17,670 homeless men in Florida vs. 9,743 homeless women. A similar disparity also occurred in Washington, with a population of 13,049 homeless men vs. 9,426 homeless women.
Moreover, the homeless rates among trans people aligned with the overall homeless proportions. By nature, the category of transgender is very broad and has a high probability of leaving out many individuals, as discussed previously, due to safety concerns and personal identity preferences. Also, It was interesting but not surprising to see certain areas, such as Georgia, having disproportionately high rates of homelessness among non-binary people since gender discrimination in the southern U.S. states is well reported.
This analysis is only a start to what could be done with this data. One potential source for future work would be, particularly once more data is available, to observe the changes between the data from 2019–2020–2021, to assess how the pandemic has affected rates of homelessness. In theory, we know that many people lost their jobs in the past year, and when more people lose their sources of income, more people will become homeless. It would be interesting to examine the actual numbers behind this — keeping in mind the limitations of the data, of course, as we touched upon above.
