UPDATE:
The LA Times made their own map right about the same time as mine (though I maintain that mine was completed first!). The data isn't exactly the same though, which I think can be explained by my conservative use of the margin of error for each ZIP code.
ORIGINAL:
After reading this LA Times article about overcrowded housing from yesterday, I decided to play around with Google Fusion to see if I could make a map that illustrated the degree of overcrowding across the country. What I came up with is a map that shows every* ZIP code in the US with over 1,000 households, color-coded to show how prevalent crowded households are in that ZIP code. (A definition of "crowded" can be found below.)
The ZIP codes are broken down into the following groups:
- Less than or equal to the national average share of crowded housing, 3.2 percent
- Between 3.3 and 6.4 percent of households—double the national average—are living in a crowded home
- Between 6.5 and 12.8 percent of households—four times the national average—are crowded
- Between 12.9 and 25.6 percent of households—eight times the national average—are crowded
- More than 25.6 percent of households are crowded
Zoom in on your city of choice to check out the areas where residents are most likely to be living in uncomfortably crowded conditions:
If you want to view the map in a full page, click here.
When I read the above LA Times article, I found the following passage incredible, and it's what inspired me to look into this further:
Cano and her family live in one of the most crowded neighborhoods in the country. Nearly 45% of the homes there are considered "crowded" — having more than one person per room, excluding bathrooms, according to an analysis of Census Bureau data spanning 2008 to 2012. Almost one home in six is severely crowded, with more than two people per room.
Southern California is an epicenter for crowded housing: Out of the most heavily crowded 1% of census tracts across the country, more than half are in Los Angeles and Orange counties, a Times statistical analysis found. They are sprinkled throughout areas such as Westlake and Huntington Park around Los Angeles, and Santa Ana and Anaheim in Orange County.
If you check out the map around LA and Orange counties, you'll indeed see a large number of red and black ZIP code areas.
I don't have a particular agenda in creating this map, but I hope people will find it useful or interesting in some way. If you can think of any ways it can be improved (or know something about Google Fusion boundary tables), I'd love to hear from you. I'm looking for a more complete ZIP code boundary table if anyone knows of one, and the populations included in the table appear to be from 2001, so an update to that data would help as well. Here's a link to the Google Fusion table I used for ZIP code boundaries.
A little bit about the methodology and definitions:
The American Community Survey has a somewhat interesting definition of "room," so almost anything but a closet or a bathroom counts: dining rooms, kitchens (as long as they're not too small or part of another living space), living rooms, and bedrooms all count. My studio, since it has a separate kitchen, would probably be considered a 0.5 person-per-room occupancy. A household with two adults and three kids that included a kitchen, dining room, living room, and two bedrooms would be given a 1.0 person-per-room occupancy rating.
To calculate the share of households that were crowded, I used data from the 5-year 2008-2012 American Community Survey. I looked at the Occupancy Characteristics (ID: S2501) for every 5-digit ZIP code tabulation area and summed up the estimated share of households with occupancy ratings of 1.01-1.5 and 1.5+. To be as conservative as possible, I then subtracted the margin of error for both of these estimates, so the values you see above are the lowest possible values within the statistical threshold used by the ACS (I'm guessing a 95% confidence level). In many, probably most cases, the actual rate of crowding will be higher than you see in the map and associated table.
I limited this to only ZIP code areas with more than 1,000 households because the margin of error on smaller areas was very high, making the data essentially useless. I chose ZIP codes because census tracts were too small to have dependable margins of error, and cities and metro areas were too large to be meaningful.
*The Google Fusion table I used to represent the boundaries of each ZIP code area was not 100% complete, but most areas were included.