Misinformation has been a real issue in the #covid19 pandemic

"How To Lie With Statistics" was first published in 1954

Its simple messages are easy to understand & use in everyday life

Let's use them to look at some covid coverage đź§µ /1
Now many of you are highly sophisticated consumers of data with advanced mathematical skills. Huff knew trying to develop such skills in everyone was unrealistic

He wanted to put us on our guard for basic things that could be picked up by simply looking /2
Here's a plot that was widely shared, showing that in Nov 20, around the time of the 2nd peak, critical care occupancy was allegedly fine

Now comparing percentages of surged ICU occupancy to historical numbers is meaningless, but what else is wrong here? /3
Well, look at Huff's discussion of figures like this

Ignore the specifics in his example & concentrate on the plot in the tweet above: changing the width of the bars deceives you

By all means ascribe no malice to the creator, but the minimising effect is real. /4
Here's another nice plot presented by a media organisation to show that those experts were way off with their modelling

Well um trends change, especially in infections capable of exponential spread /5
Huff has a relevant section for the plot above

In a fast moving pandemic "everything else refuses to be equal", presenting modelling done using a set of circumstances when they’ve clearly changed is gaslighting the data, intentionally or not /6
Next, we all look for vested interest but Huff gives us another concept, "the OK name"

#covid19 has abounded with this:
@pfizer a company of >88k employees, the @RCPath, a body that gives out exams but isn’t responsible for regulating members. These are meaningless ok names /7
No discussion of the “ok name” would be complete without the Great Barrington Declaration, falsely legitimised by the appearance of various university's names associated with signatories, when maybe the declaration’s sponsor was the most important thing... /8
These are just some examples I had to hand/remembered, maybe I'll add some more in the future. For now, remember Huff's 5 simple rules for thinking about a statistic or anything purportedly data driven /9
You can follow @toates_19.
Tip: mention @twtextapp on a Twitter thread with the keyword “unroll” to get a link to it.

Latest Threads Unrolled:

By continuing to use the site, you are consenting to the use of cookies as explained in our Cookie Policy to improve your experience.