Monday, July 20, 2020

How Do You Like Your Numbers, Sir? (I)

One of the more entertaining or debilitating aspects of following the present news is that one can find a statistic to support every argument.  Sometimes, the numbers flying past ducked heads are unreliable, inaccurate, false.  This is not necessarily an issue of bad faith.  For example:

 

WEST PALM BEACH, Fla. — The positivity rate of COVID-19 tests is a key metric for understanding how the virus is spreading, and how prevalent infections are in the community.

 

The Florida Department of Health reports the positivity rate of coronavirus tests each day -- which is supposed to show the number of positive tests compared to the number of total tests.

 

This week, the accuracy of the state's positivity reporting has come into question.

 

A review of state data shows many small, private labs have been reporting only their positive results to the state -- skewing the positivity rate higher.

 

Even after this issue came to light earlier this week, several dozen labs are still reporting 100% positivity rates, according to a review of Friday’s DOH data.

 

 

While Dr. Cole said the small labs reporting 100 percent positive tests are not likely to affect the overall state positivity rate in a huge way, it still undermines trust in the numbers.

 

“When you lose the confidence of the people, it ruins your message,” he said. “It’s very dangerous when the people don't trust what the government is telling them. It’s just a hot mess.”

 

(https://cbs12.com/news/coronavirus/dozens-of-florida-labs-still-report-only-positive-covid-tests-skewing-positivity-rate)

 

No kidding?

In all likelihood, there’s some sort of communication error going on here—some data entry peon (that’s not a slam; that used to be me) got a confused and confusing email from her boss (I’ve had these bosses) that did not make it clear she was supposed to enter ALL the test results.

But the result is that the state’s overall results are skewed—abiet “not likely … in a huge way.”


No comments: