Blog article
See all stories »

Three Ways To Lie With Charts

Lying with charts is the natural progression from How To Lie With Big Data and Beware Of Committing Harakiri By Lying With Big Data.

I've come across the following three ways of doing that.

#1. SLIPPERY SLOPE SLIDE

There are charts that gyrate up and down for no apparent reason other than that they were probably caught in a slippery slope.

Here's an example of such a chart (Exhibit 1).

EXHIBIT 1

Although the price of the company's share was INR 256.3 on both the first and the last day of the period covered by the chart, the chart somehow takes a nosedive.

Likely the handiwork of some pump-and-dump operator?

#2. FLOATING ORIGIN FRAUD

Darrell Huff showed how to mislead people by changing the origin of charts in his 1954 classic How To Lie With Statistics.

Although the book is nearly 60 years old, this tactic is widely used even today.

Fortune recently published the following chart (Exhibit 2) on the new trend of people disconnecting cable and getting all their content from streaming sites like Netflix, Amazon Prime, Hotstar, et al.

EXHIBIT 2

Anyone taking a quick glance at this chart would totally believe the article's claim that "cord cutting" - as this practice is called - is killing cable TV.

However, this is far from the truth. The chart conveys the misleading impression by shifting the origin of the vertical axis from 0% to 78%.

If we shift the Y-axis origin to 0%, which is where it belongs, we get a different picture (Exhibit 3):

EXHIBIT 3

Visually, the impact of of cord cutting on cable TV appears small.

For the quants among you, 82.1% of US households had cable TV five years ago. 79% of them still did so in 2014. The drop in cable TV penetration is 3.1 percentage points over five years i.e. CAGR of -0.77% p.a. That's hardly life-threatening by any stretch of imagination.

Let's place the two charts side-by-side (Exhibit 4).

EXHIBIT 4 

This gives you a good feel of the degree of misleading carried out by the original chart.

#3. DUAL AXIS DECEPTION

This is a relatively new tactic.

I saw its first example in this tweet, in which a cryptoenthusiast makes the following "quick comparison of Bitcoin and Fiat ATMs" (Exhibit 5).

EXHIBIT 5 

To a casual observer, it would appear that the number of Bitcoin ATMs would exceed the number of Fiat Currency ATMs by ~2019.

Obviously, that's not true.

The chart misleads readers by using drastically different scales for the vertical axis on the LHS (Fiat ATM Count) versus the RHS (Bitcoin ATM Count).

Thankfully, many tweeples fixed this chart quickly e.g. @ericturnr (Exhibit 6).

EXHIBIT 6 

LOL!

Even the mainstream media is not immune to this kind of misdirection. A few days later, I saw the following chart comparing the value of Bitcoin versus Gold on Bloomberg (Exhibit 7).

EXHIBIT 7 

Thankfully, these charts have gotten lampooned on social media repeatedly. Here's my favorite spoof of the ATM chart from @Edmund_Schuster (Exhibit 8):

EXHIBIT 8

ROTFL!

----------

These are just the three ways of lying with charts that I've come across personally.

I'm sure there are more. If you know any, please share in the comments.

The kind-hearted souls among you might argue that lying is not the only explanation for the aforementioned misleading charts.

I won't disagree with them because I can think of at least one more reason.

More on that in a follow-on post (Spoiler Alert: It's one of the seven deadly sins).

2810

Comments: (1)

Melvin Haskins
Melvin Haskins - Haston International Limited - 16 July, 2019, 08:581 like 1 like

Keep telling people about the Daryll Huff book 'How to lie with statistics'. It may be 65 years old, but it remains totally relevant today. It was compulsary reading on Maths degree courses in the 1970s. No idea if it still is, but if not, it should be,