This is a https://abs.twimg.com/emoji/v2/... draggable="false" alt="🚨" title="Polizeiautos mit drehendem Licht" aria-label="Emoji: Polizeiautos mit drehendem Licht">really problematichttps://abs.twimg.com/emoji/v2/... draggable="false" alt="🚨" title="Polizeiautos mit drehendem Licht" aria-label="Emoji: Polizeiautos mit drehendem Licht"> headline that misunderstands & misleads about what& #39;s going on.

A short thread about why counting is surprisingly hard (both in COVID and elections!) and why this matters. (1/n) https://twitter.com/CTVToronto/status/1320730439959613442">https://twitter.com/CTVToront...
You probably already see the mistake @CTVToronto made here: confusing day-to-day variation (noise) vs. reliable data (signal).

Weekend vs. weekday, for example, can influence # of people seeking tests, # of tests processed, and even the behaviours leading to spread itself. (2/n)
As @JPSoucy points out here, paying attention to the rolling 7-day average can help to smooth this out and give a more reliable picture of what& #39;s happening.

Less focus on single day values and more focus on smoothed trends = better. (3/n) https://twitter.com/JPSoucy/status/1319047536573820937">https://twitter.com/JPSoucy/s...
To @SeanDavidson_& #39;s credit (who probably didn& #39;t choose headline!), rising 7-day average is mentioned as an aside in the article... but it& #39;s the real story here, not a one-day drop.

So is the province processing about half of its own testing goal, which creates the drop. (4/n)
The bottom line, though, is we need to be patient when counting complex things. Better metrics (like cases by onset/infection date) take a much longer time to finalize. While 24hr tests are a sexy headline, they& #39;re a much less useful source of data. (5/n)
As always, @AshTuite is right on the money too: We have to pay a lot of attention to testing methodologies, like who can get a test, how easy/accessible it is, etc.

Again, focusing on the 24hr report number alone to imply patterns is /bad/. (6/n) https://twitter.com/AshTuite/status/1320092964853526530">https://twitter.com/AshTuite/...
This is a skill we& #39;re going to need to practice for November 3rd, too. Much like counting COVID cases on a massive scale is challenging, so too is counting votes. It requires a /huge/ institutional apparatus (which @clarkamiller has written about) and lots of time. (7/n)
If you want some insight into why this is super challenging - and why it& #39;s so important to have good institutions! - see this great thread by @VaughnHillyard about how mail-in ballots are counted in AZ. It involves a ton of work. (8/n) https://twitter.com/VaughnHillyard/status/1319415787807858689">https://twitter.com/VaughnHil...
There& #39;s a huge system designed to carry out each step of the process (opening envelopes, checking signatures, resolving disputes) with representatives from both parties working together to uphold democratic values in a careful, slow, methodical way. (9/n)
You can follow @ericbkennedy.
Tip: mention @twtextapp on a Twitter thread with the keyword “unroll” to get a link to it.

Latest Threads Unrolled: