Journalism: How to make the curve look much worse

Journalism: How to make the curve look much worse

Spread the love

Johns Hopkins runs a web site tracking Covid-19 world wide. The yellow curve at right illustrates what exponential growth looks like. This is bad.

I just saw a national news outlet (I won’t bother to say which one), which displays this page in a squeezed format. This is easy to illustrate with any browser – go to the web site and resize your browser window. You can get results that look like this. Check out that yellow graph line now.

This media outlet appeared to have done this intentionally, to exaggerate the yellow curve. It is scary enough as is – why make it look even worse?

The Covid-19 situation is bad. Media continues to focus on the most extreme presentations. is apparently America’s favorite shock news/hysteria/exaggerated news headline web site:

We rely on models to create projections. A model is an hypothesis about how we believe a system works. The output of a model is itself an hypothesis. Models are hypothesis generators.

We plug in scenarios and assumptions (also known some times as parameters) to evaluate possible outcomes – and to perform sensitivity analysis. In the latter, we adjust one parameter while holding other values constant to see what impact this one parameter might have.

Many of these models are used to generate a wide range of possible outcomes. These models are not necessarily making predictions about the a future – they are used to evaluate potential scenarios.

Science Magazine posted a column yesterday on the issue with use of models in our current disease outbreak. They note there are numerous unknowns with Covid-19 whose values are being estimated – consequently, the models, while useful for planning purposes, are not typically useful for making public predictions. The news media, however, seems to love the scariest model projections and runs those in a way that most people interpret this as the most likely, most plausible outcome.

Last week, Gov Newsom of California said that 56% or 25.6 million people in California would contract Covid-19 in the next 8 weeks. A few hours later, his spokesperson backpedaled and added this was a model projection that assumed no steps were taken anywhere in California – like no social distancing, no business shutdowns. In other words, this projection was a sci fi fantasy projection not based on what was actually being done in California. But all the media ran with the very frightening 25.6 million figure.

The news media has become an anxiety generator.

By the way, how many people normally die in the U.S.?

In 2014, about 2.6 million people died in the U.S. or about 216,000 per month. If 2014 is typical, we would expect about 216,000 deaths per month or 866,000 deaths in 4 months. The headline does not tell us if this 80,000 is in addition to the normal rate or an overlap. If an addition, then this would be an increase in deaths of about 9% – however, there is necessarily an overlap (think about it) so this figure is an upper bound.

Six months ago, I could have written a 100% accurate headline: American deaths will top 900,000 in next 4 months.

So how do we interpret the scary projection of 80,000 deaths in the next 4 months?


Leave a Reply

Your email address will not be published. Required fields are marked *