How to lie with statistics

[pjclarke]

Upthread.

Graphs with y-axes that don't start from zero always annoy me. It makes a small change look like a much more dramatic doubling, or something. Has anyone ever seen a climate change graph starting at zero kelvins? The resulting near enough straight line would have made it much harder to convince anybody of anything.

[EternalOptimist]

oh dear.

let's give it a rest. just for one week over christmas

[pjclarke]

What do you do if you're a sea-level rise expert denier and you want to show a graph of sea level rise without the inconvenient upwards slope?

Simples! Just rotate the graph on the page until the trend becomes zero...





(And this wasn't the blog of some washed-up TV weatherman, this was testimony to the House of Lords )

[Sysman]

10% of road accidents are caused by driving after alcohol consumption.

That means we could cut road accidents by 90% by eradicating sober driving.

True story.

Smoking kills.

A neighbour fell down the stairs on his way to buy some cigarettes.

Went to hospital for a checkup but unfortunately died later that night

[RSoles]

And of course, nearly half of the population is of below average intelligence.

[stek]

10% of road accidents are caused by driving after alcohol consumption.

That means we could cut road accidents by 90% by eradicating sober driving.

RIP Dave Allen....

[OwlHoot]

Graphs with y-axes that don't start from zero always annoy me. It makes a small change look like a much more dramatic doubling, or something. Has anyone ever seen a climate change graph starting at zero kelvins? The resulting near enough straight line would have made it much harder to convince anybody of anything.

Fancy 3D-effect graphs with an oblique "partly sideways" view (see my post above) can increase the apparent steepness even more.

[VectraMan]

Graphs with y-axes that don't start from zero always annoy me. It makes a small change look like a much more dramatic doubling, or something. Has anyone ever seen a climate change graph starting at zero kelvins? The resulting near enough straight line would have made it much harder to convince anybody of anything.

[OwlHoot]

A great book, but has anyone ever come across an example close up and personal ?

I knew a guy who would put his slides on the OHP at an angle to make the flat slope look like it was going up. A Scottish service manager who produced his stats including bank holidays from Glasgow or edinburgh, which were different. So he would get a 5% boost more often than the English guys.

Nothing Tesco-esque, but definitely telling porkies via the numbers

Saw a blatant example a few years ago - A bar graph where the taller bars were also wider in proportion, and seen obliquely so they looked much bigger by volume than their height difference alone would indicate.

Having recently bought Would They Lie to You, I'd probably recommend it. But would I lie to you? I haven't yet got round to reading it.

[EternalOptimist]

10% of road accidents are caused by driving after alcohol consumption.

That means we could cut road accidents by 90% by eradicating sober driving.

heh heh.

and 75% of traffic collisions are speed related
the other 25% are caused by cars not moving

[TestMangler]

10% of road accidents are caused by driving after alcohol consumption.

That means we could cut road accidents by 90% by eradicating sober driving.

[MarillionFan]

1 out of 10 people think I'm funny

EternalOptimist.

Which we all know is a lie. It's zero.

[BolshieBastard]

Im all for making life simpler, why complicate things with stats? Just ******* lie!

[EternalOptimist]

A great book, but has anyone ever come across an example close up and personal ?

I knew a guy who would put his slides on the OHP at an angle to make the flat slope look like it was going up. A Scottish service manager who produced his stats including bank holidays from Glasgow or edinburgh, which were different. So he would get a 5% boost more often than the English guys.

Nothing Tesco-esque, but definitely telling porkies via the numbers