Connecticut 2020 Presidential Vote and Vaccination by Town

This is a working document as I try to develop a useful way to display vaccination rates by town. Once I get it into a useful state, I plan to publish it on my blog. In the meantime, I’m going to use this as something like a whiteboard so that I can show my work while I’m still working on it.

The plot below shows the relationship between vaccination rates and percentage of the vote for Trump in the 2020 election.

Note that for the trend lines in the plot I have eliminated Mansfield and Simsbury as outliers. I think there may be an issue with Mansfield and prisons. I don’t know what’s going on with Simsbury.

The CDC uses someting they call the social vulnerability index, and DPH has included that as well. DPH flags each town that has at least one census tract that has a social vulnerability index of 75% or greater. Towns showing the social vulnerability index flag tend to have a lower vaccination rate (and a lower Trump percentage), and that’s especially true for the large cities.

Work by Charles Gaba on Trump Vote and COVID vaccination and death

Charles Gaba has been doing analyses showing the relationship between 2020 vote and vaccinations and deaths using county to level data for all states.

From Charles Gaba Which reminds me, why have I been putting Trump Vote on the y to axis? The x to axis makes much more sense given that the hypothesis is that Trump vote affects vaccination.

Polling on Partisanship and Vaccination

See this summary of the MULawPoll.

Some Quick Reactions

The first interesting thing is that the relationship between Trump vote and vaccination is so different for the over 65 crowd. It’s only the under 65 population for whom a higher Trump vote is correlated with lower vaccination (and that relationship is probably even clear if one separate out the towns flagged with the social vulnerability index or the large cities).

I’ll sleep on this and see what I think tomorrow.

Sources

Vote stats come from The Secretary of State.

Vaccination rates are from DPH via data.ct.gov.

The Issue of the Population Denominator and Its Effects on Vaccination Rates

The Department of Public Health (DPH) provides vaccination rates by age ranges by town. For this plot I have collapsed the age ranges into four groups. DPH also provides an estimated population for each range. Note that you see a lot of vaccination percentages above 100%. That’s probably an issue with the estimated population. DPH includes their estimate of the population with all of their statistics. See [this page](https://portal.ct.gov/dph/Health to Information to Systems to to Reporting/Population/Population to Statistics for this description of technical issues. I am uncertain where that leaves things. It’s possible they are using estimates based on the 2010 to 2014 surveys from Census. I am in my own town of Guilford the population is definitely aging. It became more of a bedroom community for New Haven in the 1970’s and that original influx has benn growing old together. Something similar may be going on in other towns, and that might make the estimates of population for older ages come out too low. I’m not sure. (John Burn-Murdoch of the Financial Times has pointed that that one needs to watch out for issues with the denominators while comparing vaccination rates.)

Vaccination Rate Depends Both on the Population Denominator and the Vax Doses

As near as I can tell (and I may be wrong), DPH is using the population figures from the Census ACS from 2010-2014 (a five year average). The most recent ACS data I could fetch was for 2015-2019. I used that data to estimate the population by age by town. The plot below compares the effect of using either the DPH population estimate or the latest ACS estimate as the denominator when calculating the vaccination rate. It makes a noticeable difference. The town names are show for cases where the difference in population estimates is greater than 5%.