Stop, Look, Look closer

It is easy for readers to accurately interpret graphs, despite the prevalent manipulations.

Graphs, graphs, everywhere…

There was a time when only scientists, engineers, doctors, and economists used graphs. Not any more. Anywhere you look these days, you have graphs staring at you. Newspaper, Television, Magazines, Websites, and Social Media.

This is great for readers, as they can comprehend the information in an easy and memorable manner. But, this also means that all readers must develop ‘graph interpretation skills’.  You may argue that this should be the responsibility of the people creating the graphs. It is. Yet, there are instances of authors misleading their readers with graphs. Intentional or otherwise. You, as a reader, must educate yourself to protect your interests in these situations.

A cautionary tale

Let us walk through an example of how misleading graphs can cloud your opinions and decisions. Assume that a company called MagicHealth approaches you for investing in their company. The company is  launching services for patients with chronic diseases in Canada. Your decision depends on the estimate of number of patients who would buy the services of MagicHealth.

Let us now walk through a series of graphs presented to you by three analysts. The over-enthusiastic newbie, the over-smart manipulator, and the truth-seeker. All are presenting the same data to you. Observe your thoughts and conclusions as you walk through these examples.

These examples are created using data from Canadian Chronic Disease Surveillance System.

The fancy graph

Our over-enthusiastic newbie discovered the fancy graphing capabilities of Excel. He applied it to the data and produced the graph below.

The Fancy Graph
Figure 1: This ‘fancy” graph intends to impress visually and creates an overly positive view about the potential. It shows more than 14 million people suffer from chronic diseases, the trend is increasing rapidly, and there was a sharp rise in 2007.

Take a look at the graph above and think if you agree with the following conclusions:

  1. There is a rapid rise in chronic diseases over the last 15 years
  2. More than 14 million people suffer from chronic diseases
  3. There was a sharp rise in 2007

Chances are you agree with all three conclusions. They are all incorrect. So, what went wrong? Here’s the explanations:

  1. There seems to be a rapid rise as new data was added in 2007 for two diseases – Gout and Osteoarthritis
  2. Stacking the diseases seems to show that more than 14 million people suffer from chronic diseases. This is not accurate as the same person may have more than one disease
  3. The rise in 2007 is only due to the addition of two diseases

That was about accuracy. How about interpretability? Now try to answer these questions:

  1. Which diseases are the most prevalent?
  2. Which diseases are on the rise?
  3. How many people reported mental illness in 2015? Heart Failure?

These may be the questions on you mind but it is difficult to answer them from the graph. So, the graph is not only inaccurate but also does not answer your questions.

The manipulated graph

It’s time to introduce you to the over-smart manipulator. His aim is to prove to you how lucrative the business is. I am sure the graph below will convince you.

Figure 2: The Manipulated Graph
Figure 2: This ‘manipulated’ graph exaggerates the business potential. It shows that all the diseases have high prevalence between 8 and 10 million, which is increasing during the last 15 years.

Looking at the graph, you probably concluded the following, all of which are again incorrect:

  1. All the diseases have high prevalence
  2. The prevalence is between 8 million and 10 million
  3. The trend is increasing across the years

The reality is very different as you can see in the third and final graph. So, what had the analyst done here? He has used one of the oldest tricks in manipulation – changed the scale of the axis. The y-axis is charted in logarithmic scale. This means that the distance between 0 and 1 is the same as between 1 and 10. He has also hidden the numbers on the y-axis to camouflage this. See the image below with all the numbers on the y-axis.

Figure 2a: The Manipulated Graph
Figure 2a: The trick used in the ‘manipulated’ graph is revealed in this graph as we see the complete y-axis with the labels. It is in the logarithmic scale, which compresses the difference between small and large values.

The truth seeker’s graph

Finally, let us look at the graph that aims to tell the truth. See if you get different answers to the previous questions. Also, check if all the questions from the first chart are answered:

  1. Which diseases are the most prevalent?
  2. Which diseases are on the rise?
  3. How many people reported mental illness in 2015? Heart Failure?
The Truth Seeker's graph
Figure 3: A truthful representation of the data shows the business potential. It differentiates between diseases based on prevalence and gives a sense of the growth rates too.

I am sure you can get answers to all your questions based on this graph. In addition, you can make the following additional conclusions about the top three diseases:

  1. Mental illness is most prevalent and increased from 4.63 million 5.43 million
  2. Osteoarthritis was tracked from 2007 and is on the rise. It is now the second most prevalent disease at 3.81 million
  3. Mood and Anxiety is the third most prevalent disease and has marginally increased from 3.43 million to 3.61 million

Curious to find out more?

I hope this post aroused your interest in finding more about graph manipulations. You can find out more at the following resources:

Ready to interpret

As you experienced in this post, it is prudent not to jump to interpretations and conclusions too soon. Take your time – Stop, Look, Look closer, and finally draw your conclusions. It will help you guard against misleading graphs – intentional or otherwise.

Data Source: