Voting Republican is bad for your health

Or, How not following the science has caused unnecessary loss of life in Republican states

The title is a bit flippant, but seems like an apt description of the findings on the effects of masking and vaccination on Covid-19 death rates at the national level.

This all started when I wanted to see whether mask and vaccine mandates had any effect on the death rates from Covid-19. I wanted to focus on death rates since the number of cases is no longer a useful number because of the effect of vaccinations and the lower mortality of some of the Covid variants.

Finding a good source of the various mask and vaccine rules across the country seemed like a daunting, if not impossible, task. So I used a proxy metric: how heavily Republican a state is, measured by the difference in vote percentage for the Republican nominee versus the Democratic nominee in the 2020 US Presidential election. The theory was that, the larger this difference was, the weaker the masking and vaccination rules, and the lower the vaccination rates.

If you’re in a hurry, jump to Figure 4 at the end to see the final results and conclusions.

Stage 1 — Initial results

Figure 1: Deaths per 100K vs difference in 2020 US Presidential election. Each dot is a different US state.

The initial results were not promising. There doesn’t seem to be too much of a pattern or correlation between election difference and deaths per 100K. There is a bit of an upward slope but nothing that looks too strong.

Stage 2 — The trend

Figure 2: Same as data in Figure 1, with a trend line added. Also, CA and TX are highlighted.

Plotting a trend line through the data shows that there is a correlation between the election difference (Rep % minus Dem %) and the number of deaths per 100K of population.

Since California and Texas are the two largest states (39 and 29 mil, respectively), they are highlighted to see where they fit, and they seem to be following the general trend line.

The slope is around 1.35, which means that for every 10% difference swing towards Republicans about 1,350 more people died per 10 million population. This may at first not sound like a lot of deaths, but (a) it’s on the order of the number of people who died in the 9/11 attacks and (b) scaled up to the entire country of 330 million it becomes a lot more people that died due to lax Covid policies.

Overall, the slope seems to be somewhat low, so there must also be other factors that contribute to death rates being so much higher in some states than others. One factor is population density: the denser a population is, the likelier people are of interacting with large numbers of people, increasing the transmission of Covid. This is what the next section looks into.

Stage 3 — The effects of population density

Figure 3: Same data as above, but with states’ population density represented by the size of each bubble. The three colors are for high, medium, and low population densities (yellow, blue, and red dots).

By visualizing the population density of each state, we can see a clear pattern emerging.

Figure 3 divides states into three groups:

• High population density: density above the 75th percentile
• Medium population density: density between the 25th and 75th percentiles
• Low population density: density below the 25th percentile

Looking at each group separately (yellow, blue, and red dots in Figure 3), we see a clear upward trend of death rates vs Rep/Dem election differences. This is especially true for the medium population density group.

The following section quantifies the trend for each group.

Stage 4 — Trends for each population density group

Figure 4: Same data as in Figure 3, with trend lines added.

Figure 4 adds trend lines for each population density group, and we can clearly see how strongly upward the trend line is for medium population density states.

The slopes for these trend lines are shown in the table below

The results make sense:

• In high population density areas, people frequently interact with large numbers of people, and often in tightly packed closed spaces like subways in New York, New Jersey, and Boston. In such environments, masking and vaccination rules don’t have as much of an impact. This can be because, for example, masks offer good but not perfect protection, and in extended indoor interactions in crowded areas, the viral load can get high enough to cause problems.
• On the other end of the spectrum, in low population density areas, people don’t interact with large numbers of people often, so masking and vaccination rules don’t have as much of an impact.
• In medium population density areas, the frequency of interaction between large groups of people and the level of protection offered by masks and vaccines is at the sweet spot where the masking and vaccination policy makes a large difference in how many people get sick and die of Covid. In these areas, the size of the groups you meet and the duration you meet them seems to be at the point where, for example, masking provides sufficient protection from a high viral load.

Bottom line:

In all population density groups, following the science and adhering to masking and vaccination guidelines is always correlated with a lower death rate from Covid. The effect is very strongly pronounced in medium population density areas, but is also present throughout the country.

P.S. It should be clear that the Rep-Dem % in the 2020 elections is simply a proxy metric for the presence and adherence to masking and vaccination rules. An increase in Rep-Dem % difference doesn’t directly cause more deaths from Covid. The hypothesis is that less strict masking and vaccination mandates are correlated with a larger Rep-Dem % difference, which this post has shown to be correlated with higher death rates from Covid.

Sources:

1. 2020 United States presidential election results
2. Population densities of US states
3. Covid-19 death rates by US state