Politics explain little in state-level differences in new COVID-19 cases and deaths
By Kent R. Kroeger (June 23, 2020)
A prominent, national sports reporter once told me the dumbest athletes in professional sports are tennis stars.
I hate to pick on a person when they’re down (actually, I don’t), but the story of how Djokovic most likely contracted the virus is what I find dumbfounding.
As reported in The New York Times, Djokovic and his wife, Jelena, tested positive for the coronavirus after a tennis tournament he organized in which “no one wore face masks and social distancing wasn’t enforced in the stands during the series.”
According to the Times story:
“Players mingled freely with each other after matches and posed for photographs with ball kids and tournament officials. There was no systematic testing done for the coronavirus on the participants before the event began, according to the organizers. Besides the Djokovics, at least three prominent players have tested positive: Grigor Dimitrov, Borna Coric and Viktor Troicki, a Serbian whose wife, Aleksandra, also has tested positive along with two coaches. That has prompted fears among the authorities in Croatia and Serbia that the athletes may have triggered a new wave of infections.”
Before we criticize Djokovic, consider what is happening in our own country.
Also from The New York Times on June 18th:
“Hospitals in Arizona have been urged to activate emergency plans to cope with a flood of coronavirus patients. On Saturday, Florida saw its largest single-day count of cases since the pandemic began. Oregon has failed to contain the spread of the virus in many places, leading the governor on Thursday to pause what had been a gradual reopening.
And in Texas, cases are rising swiftly around the largest cities, including Houston, San Antonio and Dallas.”
The coronavirus narrative now dominating the national media says that states (mostly Republican-dominated) where lockdowns have ended and social distancing practices are not widely practiced are experiencing a surge in new coronavirus cases and deaths.
“Nearly half of the states in the USA report a spike in new coronavirus cases, causing concern among health officials as the majority of the country implements phased reopenings.
Oklahoma is one of the 22 states with an increase in daily caseloads as officials debate safety measures for President Donald Trump’s campaign rally Saturday in Tulsa. Florida, Texas and Arizona have seen the sharpest spike.
Florida had another record day Tuesday with 2,783 additional confirmed cases of coronavirus, the largest single-day increase, pushing the state’s cumulative count past 80,000.”
These surge numbers are true and disconcerting, but as is typical with the national media’s coverage of the coronavirus, they miss the real story.
Unless one account’s for the many factors outside of the control of the governors and health officials in these “surging” states, such as population density, one is essentially spreading misinformation about what is behind these second wave surges.
A state’s population density differentiate states on COVID-19
In reality, the dominant factor associated with the past month’s increases in new U.S. COVID-19 cases remains a state’s population density (see Figure 1 and the standardized coefficient column). That factor has been behind the state-level variations in coronavirus cases since the beginning of this pandemic and it is not something any governor or state legislature can control — which may be why the news media seems to ignore its role. It’s hard to blame Donald Trump for a state’s population density.
The second most important factor in new COVID-19 cases since May 15th is whether a state is an island. In fact, there is only one state that fits that description — Hawaii — and it continues to be a shining star in the effort to stop this virus. [It helps to be an island. Just ask Iceland, New Zealand and Japan.]
Figure 1: Linear model of new COVID-19 cases within U.S. states between May 15th and June 21st.
But the governors are not off the hook — particularly Republican ones. The third most significant factor behind the last month’s increase in COVID-19 cases is the lag time (in days) between a state’s first confirmed COVID-19 case and when a state implemented a statewide lockdown policy. There is little question — waiting too long to impose the initial statewide lockdown increased the number of COVID-19 cases within a state.
Lockdown early and the decision to re-open the economy becomes much easier.
The other two factors significantly associated with new COVID-19 cases since May 15th are: (a) the percentage of the state’s population over 65 years old and (b) the change in the relative number of COVID-19 tests within the state. States with a high percentage of older citizens tended to have fewer new COVID-19 cases — likely a function an awareness that protecting our seniors is among our highest priorities during this pandemic. When it comes to testing, one reason we see large increases in new COVID-19 cases in states like Florida, Georgia, North Carolina and Texas is that they are doing more testing than ever.
That should be viewed as a good thing.
But instead, CNN, MSNBC and others in the national news media pump a flawed narrative that the problem is a bunch of fiercely ideological Republican governors and President Donald Trump promoting a ‘business-as-usual’ agenda.
If it were only that simple.
California, Oregon and Washington are experiencing these surges too — and they are controlled by Democrats.
Partisan critiques of state-level coronavirus policies are not just inadequate to explain the recent surges — they are dangerous in that they make us think we could be in control of this virus.
The impact of public policy on the spread of the coronavirus is marginal, at best. Until a reliable vaccine is developed and deployed, we are all sucking swamp water trying to stop this rude pathogen.
This is not, however, a diatribe against lockdowns or strict social distancing practices — quite the opposite, I believe those policies are fundamental to controlling this virus.
But they have limits, and before we kill the American economy in an attempt to save it, we must have a rational discussion about the policies that work and those that make minor or insignificant differences.
As this worldwide pandemic persists, I am increasingly convinced that strict social distancing practices are the key to controlling and eventually stopping this virus.
Japan (see Figure 2) and South Korea (see Figure 3) controlled the spread of the coronavirus without ever implementing lockdown (or ‘stay-at-home’) policies. New Zealand (quite effectively, see Figure 4) and Sweden (not as effectively, see Figure 5) have done the same.
Figure 2: New COVID-19 cases in Japan over time
Figure 3: New COVID-19 cases in South Korea over time
Figure 4: New COVID-19 cases in New Zealand over time
Figure 5: New COVID-19 cases in Sweden over time
What was their secret sauce? A set of cultural norms that compelled their citizens — without requiring draconian government measures — to isolate themselves when sick and to practice prudent social distancing behaviors during their day-to-day activities. Japan didn’t even need a massive testing or contact tracing efforts to stop the coronavirus spread.
We should care most about the relative number of COVID-19 deaths
It is understandable that the media focuses on the number of new COVID-19 cases since states have loosened their lockdown policies (if they existed at all).
But the outcome measure of greatest importance is the relative number of COVID-19 deaths.
Figure 6 shows the linear model estimates for new COVID-19 deaths from May 15th to June 21st. Not surprisingly, the relative number of COVID-19 cases in a state is the single most significant predictor of new COVID-19 deaths (standardized coefficient = 0.655).
More cases equals more deaths. Not a complicated equation.
Figure 6: Linear model of new COVID-19 deaths within U.S. states between May 15th and June 21st.
However, like the linear model for new cases, the relative number of new COVID-19 deaths is also a function of population density (standardized coefficient = 0.297) and whether the state is an island (standardized coefficient = -0.228). Two factors politicians can’t control.
Other significant correlates with new COVID-19 deaths —the lag in locking down a state after its first COVID-19 case, the state’s relative number of flu deaths per year (a proxy for the quality of its health care system), and the percentage of a state’s export-import economy related to China — explain relatively small amounts of variation.
More interesting, perhaps, is that whether a state is currently in a lockdown status (California, Kentucky, New Jersey, and Oregon) is not a significant correlate with new COVID-19 deaths.
Arizona is the real anomaly in new COVID-19 cases since May 15th
The linear models summarized in Figures 1 and 3 allow us to identify states that don’t seem to fit the data very well. Number one on that short list is Arizona (see Figure 7) where our new COVID-19 cases model predicts the state should have seen 1,662 new cases (per 1 million people), but instead saw 5,687 new cases (per 1 million people) in the period between May 15 and June 21st.
Overall, our linear model of new COVID-19 cases since May 15th accounts for 64 percent of the variance across states.
Figure 7: Actual versus predicted changes in new COVID-19 cases in U.S. states between May 15th and June 21st.
On the positive side, Oklahoma, Vermont and Montana witnessed significantly fewer new COVID-19 case in this time period.
If this model tells us anything, its that partisan explanations for the resurgence of COVID-19 are inadequate.
Something significantly different is going on in Arizona (a Republican state).
But, in all fairness, since the relative number of COVID-19 deaths is the real metric we should focus on, Arizona is actually doing slightly better than expected on this outcome measure (thought not a statistically significant sense).
Figure 8 shows how well our linear model of new COVID-19 deaths since May 15th performs. It predicts 82 percent of the variance across the 50 states (plus the District of Columbia).
It is a pretty good fit to the data. There are, in fact, no significant outliers.
Figure 8: Actual versus predicted changes in new COVID-19 deaths in U.S. states between May 15th and June 21st.
The news media refuses to report this fundamental reality of the COVID-19 pandemic in the U.S. — politics explain surprisingly little of the state-level variations in COVID-19 cases and deaths.
Folks, it is population density!
That said, there are policies that can make a difference, according to the data.
Early lockdowns make a difference.
Additionally, while we don’t have clean measures of the extent to which states are following good social distancing practices, it is likely that lack of these practices is central to these recent COVID-19 surges in some states. Having personally just spent the last weekend along the Jersey shore where at least half of the people were wearing facial masks, it does not surprise me that New Jersey is not — at present — witnessing a surge in new coronavirus cases.
People in New Jersey are rather obedient, believe it or not.
But that is not the case everywhere. There is anecdotal evidence of an arrogance among some people (predominately among the young and Republicans, I fear) that the dangers of the coronavirus are over-hyped.
These dangers are not over-hyped.
I believe re-opening the U.S. economy is imperative, and if the only inconvenience COVID-19 brings to your life is having to wear a facial mask when in public, consider yourself lucky.
Datasets used in this essay are available upon request at: firstname.lastname@example.org
Or tweet me at: @KRobertKroeger1