Original Visualization:

New Visualization:

I was really curious to look at the distribution of coronavirus cases in Hawaii. I found a very simple map that displayed coronavirus cases broken down by zip codes. A lot of Hawaii is dominated by natural landscapes that have no residents. However, one thing I realized from staying in Hawaii is how non-uniform the population distribution is. This is true for plenty of states, and I was reminded of the different election map visualizations we viewed that tried to demonstrate the idea that “dirt doesn’t vote”.

My edited map still uses the same data about coronavirus cases, but I’ve included new data about the populations of the different zip codes and the relative ranks of those population sizes. This provides more context with which to analyze the distribution of coronavirus cases, as a viewer can compare the population size to the number of COVID cases in a particular zip code and calculate cases per capita. I’ve also highlighted a few of these population ranks, including zip code 96720, which contains the 11th largest population in Hawaii but only has between 6 – 30 coronavirus cases. The other zip codes with population ranks 1 – 10 all have 60+ coronavirus cases. By taking a closer look at this comparable data, new and potentially actionable information is revealed. Zip code 96720 represents Hilo, which is where I am currently staying.

Even in my new visualization, which I think provides better context, there is still a lot of missing data that could clarify specific details about each zip code and give greater context to a reading of the coronavirus case distribution. For example, zip code 96863 is in last place for population size but still has 1 – 5 coronavirus cases. It actually corresponds to a Marine Corps Base, so its residents live in very close proximity and likely also have faster access to tests. Some of these zip codes reflect areas where millionaires like Bill Gates or Beyonce have bought out sections of land, while others reflect actual urban centers. Assumptions about where these activities might be occurring cannot be intuited from just the population information. There is still a lot missing from my map.

  1. Rei Zhang says:

    Cynthia,

    I think your instinct to normalize by population is also the first step I would have thought of. I agree that it’s interesting that code 96720 has such a low proportion of cases; I wonder if demographics (more seniors/middle aged people that would have already contracted COVID-19 or are taking social distancing more seriously).

    I also think your point about “dirt not voting” is really important when we consider visualizations that have population baked in; it’s really easy to conflate land area, visually, with numerical size.