Thread by @emilytroutman

As a disaster reporter, one of my major obsessions is data. Generation of useless data is one of the top ways that aid groups and governments fail people in need. 🧵

Following international disasters, there's always a lot of graphics and spreadsheets and "dashboards." This gives the impression of SO MUCH information. But it's generally not usable.

I've written a lot about data communications over the years, particularly in Haiti and Nepal. The same problems are here in the US COVID response. And I want everyone to understand... Data obfuscation is *intentional.*

As an example, I live in Montgomery County, Maryland. A week ago, the county announced it would publish daily COVID case counts for each school, then calculate the amount of schools more than 5% positive over a 2 week period, which would trigger the school to go virtual.

That was clear, responsible use of data they already had. They generated a spreadsheet (unfortunately via pdf) but it clearly showed RED schools, YELLOW, and GREEN. The data not only was understandable, its existence would trigger a set of actions. Great.

Immediately, the data showed 11 schools in the RED. They went to virtual learning. A few days later, the number of schools with 5%+ positivity rate jumped to 126, more than half the schools in the county. fox5dc.com/news/montgomer…

So Montgomery County officials backtracked and said. never mind, we aren't doing the 5% trigger anymore. We're going to look at schools on a case by case basis. At the same time, they *stopped* generating the list of which schools topped 5%. wusa9.com/article/news/l…

Important: They didn't stop generating the data, just the unique data product with schools in red. The data product was essentially too useful. Too easy to understand. Too clear. Parents looked at the RED schools and freaked out.

Now MoCo is generating daily PDFs of spreadsheets which only show the Daily Counts of COVID cases in each school. It's no longer meaningful data. Without personally downloading the PDF, converting it, and merging the sheets... there's no context. montgomeryschoolsmd.org/coronavirus/da…

A daily count of COVID cases is meaningless without knowing how many other kids and teachers tested positive in the previous 5 to 10 days and how that compares to the total population of the school.

I converted and merged the useless data into usable data and based on the last 6 days, 90 schools in Montgomery County Maryland have 5% or more positivity rates. It took me awhile because the PDFs were intentionally not designed to do this.

Whether schools close or not isn't really my point here... absence of usable data is a form of censorship. With COVID, Americans need to ask:
+ What are we counting
+ What are we not counting
+ Can I find those numbers
+ Can I use those numbers to make decisions

And give yourself the benefit of the doubt. You are not stupid! If the numbers don't give you a *clearer understanding of reality* then the data as presented is not usable. And that's probably on purpose.

When datasets and data products suddenly disappear... beware. As Joni Seager, the feminist geographer stated, "What gets counted counts"

Add: A reader sent me this incredible analysis. She is a teacher in Montgomery County and her husband created this awesome website from the deadend data.

sites.google.com/view/mcpsactiv…

Wow his data show 82% of staff at Poolesville Elementary are positive.