Let’s start this as simply as possible:
I am not an epidemiologist, nor a medical health expert of any kind.
I’ve been curious, though, about how the states where I and my loved ones live measure up to the rest of the states in terms of how Coronavirus testing is going. Thanks to The COVID Tracking Project, we have publicly available data that helps answer that question.
Of course, the data is imperfect. It’s being reported in different manners by different states, making it somewhat haphazard to compare them. Most significantly, while most states report the number of people tested, a few report the numbers of specimens tested, which can lead to certain people being double or triple-counted as having been tested (or quadruple-counted, and so on, depending how many times they’ve been tested). But while the data should be taken in the context of imperfection, it seems, by my uneducated impression, to be accurate enough at a high level to evaluate.
What, though, should be evaluated?
Much of the reporting surrounding this data centers on two pieces of information: the first being confirmed cases, the second being confirmed deaths. Obviously, these are important things. And to be clear, because the next thing I say might give the wrong impression, I don’t dispute the wisdom of reporting on these measurements. How many people have died is the most important thing we want to know, and how many people have been sick is a close second. There are questions, though, about the true number of cases and the true number of deaths. There’s a reason antibody tests and the earliest confirmed cases are prominent news stories, and that’s that, as you may well already know, testing has been too limited for us to have a firm grasp on the true extent of the virus’s spread. Which is why above, I didn’t write “how the coronavirus response is going.” I wrote “how coronavirus testing is going.” Because by my—again, rather uneducated—impression, one thing we really want to know as states take actions that reduce social distancing is how good an idea those states have of the virus’s spread within their borders.
For each state and the District of Columbia, I chose two metrics to look at, both taken indirectly from yesterday’s COVID Project daily report:
- What percentage of the population has been tested?
- What percentage of those tested have tested positive?
The importance of the first is more self-evident than the importance of the second. Clearly, a state having tested a higher portion of the populace means it knows more accurately how many people in the state have had the virus. But the second seems important too: given how little we know about the virus as a whole, it seems fair to assume no state is testing with meaningfully more precision than any others, which is a backwards way of saying that a high positive test rate implies a larger gap between tests conducted and tests necessary than does a low positive test rate.
With that established, a few more caveats before actually going through the numbers:
One is that these are total numbers—they measure how many have been tested and how many have tested positive since testing began, not just yesterday or the day before or over the last two weeks. This is a measure of how testing has gone as a whole, not how it’s going right now.
Another is those states that are reporting specimens tested rather than people tested. My impression is that there are four such states—California, Florida, Michigan, and Oklahoma—but there may be more. If you want to dig in yourself, this page has all the COVID Project’s state-by-state notes. I don’t know how big of an impact people being tested multiple times makes on the data, but my best guess, given that the number of people tested multiple times is likely a low number, is that it isn’t a hugely meaningful impact.
A third is that just because, for example, Rhode Island has completed the most tests (relative to population) and Alaska has the lowest positive test rate doesn’t mean those two states are in great shape overall in combating this epidemic. It only means testing is going decently well in those two states compared to the rest of the country. I don’t have the data to get into how the U.S.A.’s overall testing compares to the rest of the world.
A fourth is that just because testing hasn’t gone as well in one state as another does not necessarily mean that the first state has done a better job than the second. It would be unreasonable to expect states hit early by the outbreak or states that have a more urban population to have as encouraging of testing numbers as isolated states or rural states. There’s a higher need for testing in some states than others.
Finally, I took the population data from the Census Bureau’s 2019 estimates.
Now, how each state’s completed test number compares to their population:
| State | Completed Tests/Total Pop. | One Completed Test per __ People |
| RI | 7.2% | 14 |
| NY | 5.3% | 19 |
| MA | 4.8% | 21 |
| ND | 4.8% | 21 |
| LA | 4.0% | 25 |
| UT | 4.0% | 25 |
| NM | 3.9% | 26 |
| DC | 3.4% | 29 |
| NJ | 3.2% | 31 |
| TN | 3.2% | 31 |
| AK | 3.1% | 32 |
| WV | 3.1% | 33 |
| CT | 3.0% | 33 |
| WA | 2.8% | 35 |
| VT | 2.8% | 36 |
| IL | 2.7% | 37 |
| MS | 2.7% | 37 |
| DE | 2.5% | 39 |
| HI | 2.4% | 42 |
| MD | 2.3% | 43 |
| MI | 2.2% | 45 |
| AL | 2.2% | 46 |
| FL | 2.2% | 46 |
| SD | 2.2% | 47 |
| OK | 2.0% | 50 |
| NH | 2.0% | 51 |
| CA | 2.0% | 51 |
| PA | 2.0% | 51 |
| IA | 1.9% | 52 |
| GA | 1.9% | 53 |
| WY | 1.9% | 53 |
| AR | 1.8% | 55 |
| NE | 1.8% | 56 |
| IN | 1.7% | 58 |
| ID | 1.7% | 59 |
| WI | 1.6% | 63 |
| MN | 1.6% | 64 |
| NV | 1.6% | 64 |
| OR | 1.6% | 64 |
| MO | 1.5% | 65 |
| ME | 1.5% | 65 |
| CO | 1.5% | 68 |
| TX | 1.5% | 68 |
| NC | 1.4% | 69 |
| MT | 1.4% | 70 |
| KS | 1.4% | 73 |
| OH | 1.4% | 73 |
| KY | 1.3% | 74 |
| VA | 1.3% | 75 |
| SC | 1.3% | 76 |
| AZ | 1.2% | 82 |
And, how each state’s positive test rate stacks up:
| State | Positive Tests/Completed Tests |
| AK | 1.6% |
| HI | 1.8% |
| WV | 2.3% |
| MT | 3.0% |
| ND | 3.5% |
| UT | 4.3% |
| OR | 4.3% |
| NM | 4.9% |
| VT | 5.2% |
| OK | 5.2% |
| WY | 5.5% |
| ME | 5.9% |
| TN | 6.3% |
| AR | 6.4% |
| ID | 6.9% |
| WA | 7.1% |
| CA | 7.2% |
| AL | 7.8% |
| TX | 7.8% |
| FL | 8.0% |
| NC | 8.1% |
| KY | 8.7% |
| MN | 8.9% |
| WI | 9.3% |
| MO | 9.4% |
| NH | 9.6% |
| SC | 10.0% |
| MS | 10.2% |
| AZ | 10.5% |
| NV | 11.7% |
| RI | 13.0% |
| OH | 13.0% |
| KS | 13.6% |
| SD | 14.3% |
| GA | 14.8% |
| LA | 15.9% |
| IA | 16.7% |
| NE | 17.6% |
| VA | 17.8% |
| IN | 18.2% |
| IL | 19.0% |
| MD | 19.4% |
| CO | 19.9% |
| MI | 19.9% |
| PA | 20.3% |
| MA | 21.1% |
| DE | 21.8% |
| DC | 21.9% |
| CT | 28.2% |
| NY | 31.2% |
| NJ | 45.4% |
And, lastly, a graph comparing the two:
On this graph, to be clear, the further right a state is and the further down a state is, the better that state’s understanding of the extent of the virus’s spread (roughly, of course—this is all a rough way to measure this thing, even if it’s close to the best it seems we can do as the public). It’s hard to make out a lot of individual states on here, but the outliers are the interesting ones: Rhode Island testing the most. New Jersey with the highest positive test rate. Alaska, Hawaii, and West Virginia all reporting low positive test rates. North Dakota seemingly in a comparably good place with testing.
It’s all rough. All of this is rough. But these are numbers worth examining, especially as you’re trying to evaluate the relative danger of interpersonal interactions right now. Take it all for what it’s worth.
