“How Charts Lie” by Alberto Cairo – 8/10

Source of charts from: https://www.dropbox.com/sh/65025l954m3uzok/AADCRbkrNXLjl4-jJU6FB7tCa?dl=0

“ The world cannot be understood without numbers.  And it cannot be understood with numbers alone.” – Hans Rosling, Factfulness (2018)

Principles

• “The success of any chart doesn’t depend on who designs it, but also on who reads it, on the audiences graphicacy or graphical literacy.”  – Albert Cairo
• A chart shows only what it shows, and nothing else.
• Don’t read too much into a chart – particularly if you’re reading what you’d like to read (false precision)
• Different levels of thinking may require different levels of data aggregation (country, county, city, individual).
• Any chart is a simplification of reality, and it reveals as much as it hides.
• Are the patterns or trends on this chart, on their own, sufficient to support the claims the author makes?
  • For example: Is the ACA good for the economy?  Did it improve the job market?

Guidelines for Charts

1. Baseline of a chart should start at zero (don’t distort the proportions – see below)
   1. Truncation is good if it doesn’t distort the baseline and it helps you convey what you want to say (see below)

• Most charts encode data (emphasize using different techniques)
  • Lines/rectangles/circles
  • Color
  • Position
  • Height/length
  • Size or area
• How to determine axis width: height (aspect ratio)
  • If change is 35% (increase in life expectancy), use 3:1 (length to height) – see below

How to Read Graphs/Scatter Plots (see below chart)

1. Title, Introduction (or caption) and source
2. Measurements, units, scales, and legends
   • Scale labels (x and y axis)
3. Methods of Visual Encoding
4. Read Annotations
   • Short textual explanations to charts to emphasize main takeaways or highlight relevant points (within the chart)
5. Take a bird’s eye view to spot patterns, trends, and relationships
   • Scatterplot: Look at the dots themselves and see if there is any trend lines
   • Divide the chart into 4 quadrants

Ideal Chart Example:

Charts That Lie By Being Poorly Designed – Not the same baseline

Rules of Thumb with Charts:

1. Distrust any publication that doesn’t clearly mention or link the sources of the stories they publish
2. Try a varied media diet – left, center, and right.
   • Don’t follow very partisan sources
   • Expose yourself to sources you disagree with and assume good faith on their part (that they’re not lying)
3. Follow sources that issue corrections when they ought to and that do it visibly – shows high civic and professional standards. 

Charts That Lie by Displaying Insufficient Data

• The question is: “What is the most successful movie ever?”  The chart below shows the box office weekend records.  The conclusion would be “The Force Awakens” based on what’s shown.
• What are some issues with the chart? 

Issues:

1. Dollars aren’t on the same basis (inflation)
2. There are more movie theatres nowadays than before – what if Jaws was released in as many theaters as today?  Could look at opening weekend sales
3. More production costs – try to normalize for difference in cost to determine success

By evaluating based on their return in the first weekend, “Every Which Way but Loose” could be deemed the most successful movie.  There are many ways to gauge success, but finding a way standard way to normalize is important. 

Charts That Lie by Concealing or Confusing Uncertainty

• Election polls are typically shown at a point estimate. In an PA election between Conor Lamb and Rick Saccone, Rick was leading by 3 percentage points headed into election day. 
• However, Rick lost the election! 

The chart below is a better way to show poll results – it displays a range of outcomes with confidence intervals.

Charts that Lie by Suggesting Misleading Patterns

Correlation doesn’t equal causation

Correlation Case study 1:

• Based upon the graph, the conclusion is “the more cigarettes we consume, the longer we live!”  We know that this isn’t true and there’s something driving the results…wealth.

• People in wealthier countries tend to live longer because they have access to better diets and healthcare and are less likely to be the victims of violence and wars.  Plus, they can afford to buy more cigarettes.

By separating out the high/middle/low income countries, you can see some various trends:

• Poor nations have high variability in life expectancies, but on average, they don’t smoke that much
• Middle income countries have a large variation in life expectancy and cigarette consumption and the relationship between the two variables is weak.
• High income countries tend to have high life expectancies overall, but their cigarette consumption is low or high depending on the country. 

Correlation Case study 2:

These charts below are correlated together, but does the fact that Nicolas Cage appears in more movie provokes more accidents?  Or could it be the fact that Nicolas Cage appears in more movies provokes more accidents?

Conclusion

• Rationalization vs. Reasoning
  • Rationalization is a dialogue within ourselves or with like-minded brains (confirmation bias – notice data that fit with our beliefs)
  • Reasoning – scrutinize ideas ore carefully if we don’t like them than if we do.  Need to surround yourself with different minds and opinions and understand how they got to that conclusion with their logic and reasoning. 
    • “an honest and open conversation in which we try to persuade speakers who don’t necessarily agree with us beforehand with arguments that are as universally valid, coherent, and detailed as possible, while opening ourselves to persuasion.