How to lie with Statistics - Coronavirus Special

Upforthematch

I think this is an interesting aspect of Coronavirus that needs to be discussed separate to the main thread.

Most people are now saying that China 'cooked' their figures. This could also mean that some other government statistics mightn't be worth the presentations they are written on.

Have you noted any interesting examples of how health authorities around the world are displaying statistics and how they are being interpreted?

I think this is a classic one below. Show Joe public a graph that can only ever realistically be downward sloping (%'s on the left axis) and he might be persuaded to think that things are improving or not so bad.

joe_99 wrote: »

This graph shows the trend is positive. If it was going in the other direction we'd be in a much worse position.

Incorrect reading of it?

"Downward sloping curve - therefore it's ok! :pac:"

I'm so cynical about how figures are being presented in general - not targeting DOH- that I don't think Coronavirus is going to be an exception.

Of course, any epidemic will show massive growth rate at the beginning - it is starting from a base of zero after all and when a full population is infected the growth rate is zero!

My main point is look closely at the axis label , how the data was gathered and imagine what the opposite situation would look like.

Would love to see other examples.

Harry Palmr

That's not a lie, it's a chart that many don't know how to interpret.

blindside88

I’ve tried to explain to several people how different statistics around things like this can be given in different ways to make them sound worse or better.

Example: Day 1, 2 people had the virus. Day 2, an extra 1 person is infected.

Reported as “Reported cases jump by 50% in one day”
Or could be reported as “one extra case today”.

Both are true but one sounds much worse than the other

Upforthematch

Harry Palmr wrote: »

That's not a lie, it's a chart that many don't know how to interpret.

That's the beauty of it. Knowledge of statistics in general is not great.

Is the appropriate care being taken to explain these graphs - that's what I would like to know.

It's all in the interpretation.

Upforthematch

blindside88 wrote: »

I’ve tried to explain to several people how different statistics around things like this can be given in different ways to make them sound worse or better.

Example: Day 1, 2 people had the virus. Day 2, an extra 1 person is infected.

Reported as “Reported cases jump by 50% in one day”
Or could be reported as “one extra case today”.

Both are true but one sounds much worse than the other

Absolutely.

The mean v median debate is another.

I would be very worried if I heard any government referring to mean averages instead of median for ages of individuals infected.

Or worse - simply referring to "average".

Upforthematch

Here is a nice simple one from a website called 'The Globalist' - an article from 15 March - contains the below graphic from WHO data dated 23 January.

Presenting old data in a recent article could give the impression that Coronavirus is not *that* infectious. If that is the impression you wanted to give.

But it's not a lie!! :pac:




https://www.theglobalist.com/china-coronavirus-sars-pandemic-global-health/

hmmm

There's nothing wrong with that first chart. You're reaching now trying to turn this into some sort of conspiracy.

Upforthematch

hmmm wrote: »

There's nothing wrong with that first chart. You're reaching now trying to turn this into some sort of conspiracy.

No conspiracy, I can safely say I don't think DOH or HSE are out to con people.

I should mention that "How to Lie with Statistics" is the name of a famous popular book in the area of statistics. It's really about how statistics can be presented and easily misinterpreted.

We are lucky to live in a country that has accountability but this is not always the case elsewhere - that's why it's important to look a bit closer at the data.

Beasty

All been discussed repeatedly in the main thread

A standalone thread is just going to encourage conspiracies