Image credit: NIAID at Bloomberg
With all 50 states beginning some type of reopening after Stay-at-home orders and business closures, I wanted to assess the total number of COVID-19 cases/deaths comparing population density and Governor political affiliation in a state by state basis (I previously wrote a post on how States responded to COVID-19 based on political party, here). We understand that urban centers are getting hit hard by COVID-19, and many of these areas are led by Democratic leaders. Thus, which is a better indicator of greater COVID-19 cases and deaths? Population density, or political party affiliation of the states’ Governor.
These data analyzed are from COVID-19 Tracking Project up through May 26th. Any increase in cases since then due to reopening and increased social gatherings are unaccounted for. Population data was taken from the U.S. Census Bureau. Total land area (square miles) was taken from Wikipedia. States per region was assessed using maps previously designated by the U.S. Census Bureau. Some caveats to remember: 1) correlation does not mean causation. 2) Statistics can be analyzed and interpreted utilizing many, many methods. Let’s begin!
How I organized the data
The states that are members of each respective region (Northeast, Midwest, South, and West) can be seen below. To calculate population density per square mile in each state, I took the total population of each respective state, and divided it by the states’ total land area. After that, I organized the data by region, greater/less than 100 people/mile2, or the political party of the Governor in each state. After that, I calculated the mean (average), standard error, and graphed it for display.
How to determine if your data is significant
It’s great to display cool graphs and data, but how can we ensure that the statistics and data are meaningful? We use statistics! For these data, statistical analyses were deemed significant if the probability value (aka p-value) was <0.05 (*), <0.01 (**), or <0.001 (***). The p-value is the probability that the results of your data go against your null hypothesis. In this case, our null hypothesis is that there is no difference in cases, deaths, or tests for COVID-19 in each region (based on population density) or political party. Basically, the smaller our p-value, the more likely our null hypothesis (there’s no difference) can be rejected – i.e. the greater likelihood that there population density is an indicator for the number of cases and deaths of COVID-19 in the states.
Data Time
Okay! First, we’ll look at the population density of states in the various regions (see regions above) of the United states. The Northeast has significantly higher population density compared to the other three regions of the US (Figure 1, upper left). This then leads us to find that the number of deaths and cases per capita in the Northeast are also significantly greater (Figure 1, bottom graphs).
Interestingly, the number tests being performed is the basically the same across all the regions. The Northeast tests a bit more than the other regions, but only significantly more than the west (barely). This means that the substantial more cases and deaths in the Northeast is not due to simply testing more people, at least for sure compared to the Midwest and South.
Many of the states in the Northeast vote liberal or democratic in national elections. Could political party account for the increased cases/deaths? In Figure 2, I organize the data on a state-by-state basis sorting the data to compare cases/deaths of COVID-19 between states with greater/less than 100 people/mile2. In states with more than 100 people in a square mile (26 states and Washington, D.C, 10 with Republican Governors, 16 with Democratic Governors, D.C. Democratic Mayor), there have been significantly more cases and deaths of COVID-19 than those states with less than 100 people per square mile (24 states – 16 with Republican Governors and 8 with Democratic Governors). Again, all of the states are testing at the same rate.
If we look the number of cases and deaths between states with Democratic or Republican Governors, we find states with Democratic governors had more cases and deaths of COVID-19, but not enough to satisfy statistical significance. Thus, the probability that political party of Governors accounts for differences in COVID-19 cases or deaths is low.
Final Thoughts
Again, these quick analyses are by no means the end of analyzing the impact of SARS-CoV-2 has had on the global population, let alone the myriad of other weighing factors that have affected the spread of COVID-19. For instance, one group recently analyzed and published how the shelter in place and business shutdown policies influenced the caseload in over 1700 places around the world and found that these policies reduced over half a BILLION cases of COVID-19. Further, the World Health Organization just announced that new data has shown that cases of asymptomatic transmission of COVID-19 are rarer than previously thought, and “superspreaders” might be most responsible for the dissemination of the disease. These data are important to furthering our understanding of the virus and disease, and are going to be even more crucial to gather as we continue reopening our communities.