Read on Twitter
Clarification thread (brief)This will be my only tweet(s) for Super Bowl Sunday (unless the Bucs bring it home tonight).
I've been asked why my vaccination rates or doses per 100,000 calculations for Florida don't match other national dashboards.
Answer is simple
1/
To calculate doses administered per 100,000 population...
- most dashboards take the cumulative doses administered to date *divided by* the total state population
- I have the same numerator but my denominator is different as I restrict to the population 16+ years
2/
- most dashboards take the cumulative doses administered to date *divided by* the total state population
- I have the same numerator but my denominator is different as I restrict to the population 16+ years
2/
Why?
Well, the denominator should include only people who could be included in the numerator.
Currently, only people 16+ are eligible to receive the vaccines being disseminated.
I want to reflect doses administered per 100,000 population who is ELIGIBLE TO RECEIVE THE VAX
3/
Well, the denominator should include only people who could be included in the numerator.
Currently, only people 16+ are eligible to receive the vaccines being disseminated.
I want to reflect doses administered per 100,000 population who is ELIGIBLE TO RECEIVE THE VAX
3/
Does it matter for county or state comparisons?
It can.
Example
- Assume Utah and Vermont are actually vaccinating those 16+ at the exact same rate: 20%
- A measure using their entire pop as the denom would show Vermont at 16.8% of people vax, while Utah would show 14.8%.
4/
It can.
Example
- Assume Utah and Vermont are actually vaccinating those 16+ at the exact same rate: 20%
- A measure using their entire pop as the denom would show Vermont at 16.8% of people vax, while Utah would show 14.8%.
4/
Both underestimate the % of the vaccine eligible population who has been vaccinated, but the underestimate is more in Utah.
Why?
A higher proportion of the population in Utah is 15 or younger (25.8%) compared to Vermont (16.0%).
5/
Why?
A higher proportion of the population in Utah is 15 or younger (25.8%) compared to Vermont (16.0%).
5/
I've chosen extremes in terms of highest and lowest state percentages of the pop 15 or younger, so the impact of including the entire pop in the denom would not typically result in such big diffs.
But I also compare counties in FL & age distributions can be very different.
6/
But I also compare counties in FL & age distributions can be very different.
6/
Anyhow, that's why I do it the way I do.
If the vaccine were to become available to younger people, I would obviously adjust my approach.
Also, if I'm think about herd immunity instead of the ability to vaccinate people, I might also modify the approach.
Hope this helps!
7/
If the vaccine were to become available to younger people, I would obviously adjust my approach.
Also, if I'm think about herd immunity instead of the ability to vaccinate people, I might also modify the approach.
Hope this helps!
7/
And...........................................
8/end
8/end
