[This article has two co-authors. The first is Eric Tyberg, a retired IT executive and consultant residing in Lincoln, California. Originally from Falun, Wisconsin, he rose through the ranks at IBM and formed his own consulting business when IBM downsized. The second is Peter A. Morrison, Ph.D., RAND Corporation Senior Demographer (retired) and President, Peter A. Morrison and Associates, Inc. He is an applied demographer based on Nantucket Island, MA. He graduated from Dartmouth College and holds a Ph.D. from Brown University. Email: petermorrison@me.com; website: Peter A. Morrison].
Extending our earlier work exploring the 39 counties of Washington (see article and commentary), we then explored the implications of two research findings and their implications for the U.S. as a whole: (1) Republicans are far likelier than Democrats to believe their home state hasn’t moved fast enough to reopen businesses and ease restrictions; and (2) Democrats are far likelier than Republicans to report wearing masks in public. Adding fuel to the political flames, It also turns out that conservative media misinformation may have contributed to the severity of the pandemic, which, in part, may explain Republican attitudes and behaviors.
As was the case with the article on political orientation and Washington’s counties, we acknowledge that there is a range of plausible demographic factors affecting COVID-19 transmission, infection, and reporting. These factors can include: self-selected participation at protests, presidential rallies, houses of worship or other “super-spreader” venues such as assisted care facilities. Many of these factors may correlate with people’s political orientations, but not necessarily cause or predispose such orientations.
Our logic here stems from political scientists’ reliance upon “homogeneous voting precincts” to discern the favored candidate of one or another type of voter. If 9 of every 10 eligible voters in precinct X have Latino surnames, it’s a sure thing that whichever candidate carried that precinct must have been Latinos’ top choice. By that same logic, heavily Republican or Democratic-leaning counties point to political orientations that may prompt their residents to practice social distancing and wear masks—or not.
We’ve compared the 178 million residents of Democratic-leaning counties (487 counties that Clinton carried in the 2016 presidential election) with the 150 million residents of Republican-leaning counties (the 2,657 counties that Trump carried). Here’s what we found by simply comparing these two aggregated populations:
· April 1st, 2020: Initially, residents of Republican-leaning counties registered far fewer COVID-19 cases than did residents of Democratic-leaning counties (46,356 cases vs. 159,176 cases).
· April 1st -June 15th, 2020: The initial gap eroded, as cases increased by 1,264 percent in the 2,657 Republican-leaning counties compared with only an 815 percent increase in the 487 Democratic-leaning counties.
· April 1st, 2020: Initially, the Democratic-leaning counties had nearly three times the per-capita case-rate found in Republican-leaning counties: 81 cases per 100,000 people vs. 31 cases per 100,000 people, respectively.
· April 1st -June 15th, 2020: As the case-rate gap narrowed, the 487 Democratic-leaning counties registered just under twice the per-capita case rate found in the 2,657 Republican-leaning counties; that is: 817 cases per 100,000 people vs. 421 cases per 100,000 people, respectively.
Many Republican-leaning counties across the country are sparsely populated and isolated, which likely accounts for why they registered very low or even zero case-counts as of April 1st. The 2,225 residents of Garfield County, WA, some 300 miles southeast of Seattle, where population density averages 3 people per square mile, had yet to register a single, confirmed case as of July 1st. This is one of the major reasons the April 1st case count nation-wide is much higher for Democratic-leaning counties: they are far more densely populated. Taken altogether, the 487 Democratic-leaning counties have an average of 321 people per square mile, while the 2,657 Republican-leaning counties have only 46. The denser counties also tend to be on major transportation routes, often serving as hubs.
Despite the low case counts on April 1st, these less densely populated Republican-leaning counties as a whole are now experiencing noticeably higher percentage increases in case counts than the Democratic-leaning counties as a whole. Concordantly, their per-capita case rates have also grown at a faster rate than found in the Democratic-leaning counties. This is remarkable - perhaps even ominous - when one considers the possibilities.
Could the virus now be catching up in many Republican-leaning locales—the ones that never closed down or were hell-bent on re-opening? Are their residents ill-disposed to outside mandates to self-isolate, practice social distancing, and wear masks, possibly due to the misinformation they consume from conservative media outlets? Is parental defiance of truth transmitted to children, placing them at risk? Our findings here support these views as do our findings in two “Democratic states,” California and Washington, both of which were carried by Clinton in 2016 (see article and commentary).
Our take is that political orientation should be considered along with other factors likely to generate COVID-19 cases. So, along with testing and its accuracy, data suppression, potential “super-spreader” venues, population density, rates of interaction, age, race and ethnicity, and gender, we believe that “being Republican,” or being in proximity to them, could be a very real risk factor.
Data Sources.
US_County_Level_Election_Results_08-16/2016_US_County_Level_Presidential_Results.csv
List of United States counties and county equivalents
Alaska Presidential Results by County Equivalent, 1960-2016
Comments by Readers
Michael Riordan
Jul 04, 2020I agree that political orientation may indeed be a causative factor for Covid-19 infections, but it seems hard to tease this out from the other, correlated factors. As the old saw goes, “Correlation is not causation.”
David A. Swanson
Jul 04, 2020My general impression of the readership of NWC is that it is sufficiently educated/knowledgable that defining risk factor was not needed. Now I realize that assumption is likely misplaced. Here is a definition of risk factor (from https://www.medicinenet.com/script/main/art.asp?articlekey=5377):
Something that increases a person’s chances of developing a disease. For example, cigarette smoking is a risk factor for lung cancer, and obesity is a risk factor for heart disease.
We can continue to refine the defintion of “risk” in the mamner it is used in quantiative social sicence, medical, and public health research. For example, you could look up “hazard rates,” “odds ratio” and “relative risk.” Odds ratios are particularly useful if one has a dichotomous dependent variable (which is a severe violation of one of the seven major assumptions underlying ordinary least squares (OLS) models) because by taking its log and thereby creating a “logit,” one can bring to bear OLS tools and their variants.
As the preceding discussion suggests, these quantative renditions of risk are usually seen in the context of a multivariate probabilisric modeling context (e.g., a logit model), such as found in a paper on the risk of smoking given at the 5th Symposium on the Human Mortality Database held last year in Berlin at the Free University, where “hazard rates” where employed among other probabilisitc measures
https://www.mortality.org/Public/HMD_5th_Symposium/Public/Swanson_abstract_new.pdf
David A. Swanson
Jul 04, 2020The universe around us is difficult to comprehend, as is the social universe and therefore, the public health universe. The dichotomy, “cause v. correlation,.” is a rendition of pre-reflective critical thinking (see the commentary on the NWC article, “The Second Time as Tragedy”). This type of thinking is virtually non-existent in modern quantiative social science research (including public health research). Here is an example of how “risk factor” is employed in current public health research. It is In the form of selected text from a journal article abstract, wherein “risk” is used in the maner it is employed in “Is Being Republican a Risk….”
“Whilst HLA matching is routine in renal transplantation the possible benefits of matching donor to recipient age have not been previously explored….Multivariate analysis, taking into account other variables (HLA matching, dialysis time and type, donor/recipient sex, local/imported kidneys, sensitivity, operation time, total ischaemic time, pre-operative transfusions) indicated that age difference was the single most important variable (P less than 0.01). The only other important covariate risk factor in improving graft survival was HLA-DR matching (P less than 0.05). Donor-recipient age difference is a potentially important recipient selection criterion in cyclosporin-treated renal transplant patients.”
https://pubmed.ncbi.nlm.nih.gov/1910435/
The point here should be obvious, what is it about “age” that makes it a potentially important risk factor in terms of a renal transplant? It is likely a complex of factors that will require a multivariate, probabiliistic approach to examine. Similarly, “Being Republican” appears to be a risk factor. What is it about “Republican” that makes it a potentially important risk factor in terms of covid-19 transmission? It is likely a complex of factors that will require a multivariate, probabilistic model (and the high quality data needed to feed it) to examine. It would be very satisfying if we could identify the causal mechanism, but it is more likely, given our collective ignorance and the complexity we face, that “risk factors” and their variants will continue to used for quite a while as we “naked apes” continue our explorations.
So, where does this leave us? The article upon which you commented suggests that being “republican” is a risk factor. It is based on empirical research. As we concluded in the article and note in several places within it, more definitive answers will require the mutlivariate, probabilisitc tools described. Even then, there is nothing definitive in science. It’s utlimately an unstructured question with no definitive answers, a post-reflective state of mind for which we naked apes may turn out to be ill-equipped to handle.
Tim Paxton
Jul 04, 2020Which Face Mask has FDA testing and approval?
Why does a N95 mask have a one way valve (out) so one can sneeze onto the lettuce in the grocery store?
Can Governors prescribe mandatory defective/fake medical devices (masks), not approved by the FDA, which are proven to be worthless on healthy / asymptomatic free people?
Who ever heard of quarantining the healthy?
Should Bellingham provide “Wash In/ Wash out” hot water and soap hand wash stations for all restrooms, and entryways to restaurants, bars, stores? )To help eliminate the fecal contamination route.)
Should Bellingham Council subsidize Bidets for Bellingham housing? To help keep the hands squeaky clean?
Should Nursing homes be locked down during an epidemic to protect the vulnerable populations?
If First World countries such as Sweden, which did not lock down/destroy its economy have low contamination rates, should the USA copy that model?
David A. Swanson
Jul 04, 2020Instead of lame jokes and comments that remind me of the graffiti over the urinal in the men’s room at Tony’s Tavern in Custer, let’s have a real duel, “bushi v. bushi,” - Haiku. Since Trump claims covid-19 was sent to us by the Chinese, here is my opening shot.
A pox from the east -
I banished it from the land
It haunts my events
David A. Swanson
Jul 04, 2020P.S. You need to do your research. Sweden is a covid-19 disaster. The Prime Minister is launching an official inquiry into his own (coalition) government and the decision (his) to follow the “no containment” strategy: https://fortune.com/2020/07/01/sweden-covid19-death-toll-investigation/