Numbers don’t lie – graphs do

There are basically two ways to fake a statistic: Winston Churchill’s famous quote “I only believe in statistics I doctored myself” is a reference to the first way, often used by governments for propaganda purposes.
They operate with a simple ‘fake input numbers – fake output statistics’ mechanism to mislead the unaware public to get re-elected. A poor political game that shouldn’t bother you too much.

More of a concern however should be the second, more sophisticated way, to fake a statistic, I would call it ‘Numbers don’t lie – graphs do’.

Graphs are heavily used to make information more understandable and because of their communicative effectiveness.

But charts do often depict comparisons between data or trends to produce a desired result.

Thereby, the input numbers are correct, the graphs, charts or diagrams we encounter daily in papers, magazines or online are nonetheless misleading, if you don’t know how to read it.

I like to stress that misleading graphs are not always deliberately fabricated, sometime it’s just a case of people not understanding the data behind the graph they create.

Let’s have look.

Figure 1 – House Prices
Figure 2 – Advertising
Figure 3 – Decline or Depression
Figure 4 – Burger Sales
Figure 5 – A false piece of pie
Figure 6 – Gold- & Oil price
Figure 7 – A crash disappears

Figure 1

Dramatic rise in house prices!

From this graph it looks as though house prices have trebled in one year. It’s misleading because the vertical axis does not start at 0.

The correct version of the same graph gives more accurate picture of what has happened.

Figure 2

Dear reader,
we believe you’re stupid

The most important asset of a newspaper is it’s credibility. With this graph, ‘The Times’ suggests to have twice as many readers as ‘The Telegraph’ by starting the vertical axis at 400’000 instead of 0. In fact, the difference is only 10%. Would you trust this newspaper?

Figure 3

Decline or Depression?

If the vertical axis does not start at 0, an economic slowdown looks like a great depression.

Figure 4

A very, very Big Mac

This time, the vertical axis starts indeed at 0. But the graphic representation of the sales figures of Mc Donald’s and Burger King is still misleading. The numbers don’t lie, but the use of a linear axis instead of a logarithmic gives the wrong picture of the proportions. The four times larger sales figure of McDonald’s in reality appear to be at least 10 times higher in the graphic.

Figure 5

A bigger slice of the pie

The pie chart is an easy way to convey information. Making a pie chart 3D will make a correct interpretation difficult.
Often used for aesthetic reasons, the third dimension does not improve the reading of the data. On the contrary, because of the distorted effect of a perspective with 3D.

Figure 6

Linear charts (left) pace different levels on the vertical axis equidistantly.

The increase in the gold price (yellow) from 400 to 500 (plus 25 %) gets the same distance on the linear chart as the move from 1400 to 1500 (plus 7%) thus distorts the picture.

A Logarithmic chart (right) spaces the levels differently to reflect the percentage size of the move.
An increase from 400 to 800 gets the same distance on the vertical axis as an increase from 800 to 1600.

Figure 7

A mayor crash disappears

Even if you only have little knowledge about the history of the stock market, you might have heard about the devastating crash of 1929 that triggered a global economic depression (see chapter 29).

In a linear chart showing the historical performance of the Dow Jones (top), this mayor event simply disappears, whereas a logarithmic chart displays the magnitude of this crash accurately.

Next chapter: http://www.theleader.ro/license-to-pay/