Making Charts Say What You Want Them to Say

July 10, 2008

Check out the following two charts:

Both charts show the daily values for the DJ Financials Index. And, if you could see the actual numbers, you would see that both charts show that the index is down around 30% in 2008. The only difference is the scale of the axes between the two charts. The first chart makes the performance look a lot worse because the axes scale is a lot smaller (maximum value of 750 and minimum value of 300 on the left axis). The second chart has a much broader axes scale with the left axis showing a maximum value of 1,000 and a minimum value of 0.

What’s the point of this? Well, you have to be aware that authors can easily make something look better or worse than it really is just by altering the numbers. Whenever I read an article with an accompaning chart, I ALWAYS look at the scale of the axis so that I can assess whether or not the author is trying to manipulate the numbers.

Just remember: things may not always be as bad or as good as they seem.

UPDATE: I decided to go ahead and make the Excel spreadsheet I used for this post available for download.

11 responses to Making Charts Say What You Want Them to Say

  1. I had a stats professor in college say, “Figures never lie…but liars figure.”

  2. “Well, you have to be aware that authors can easily make something look better or worse than it really is just by altering the numbers.”

    … or rather, *WITHOUT* altering the numbers.

  3. There is a good book that outlines some of the key ways to do this called How to Lie with Statistics (http://en.wikipedia.org/wiki/How_to_Lie_with_Statistics)

  4. Ok.. is it just me, or is it not just the scale that is different, but the numbers for the Daily values are different. Scale wouldn’t adjust the shape of the pattern, and it’s gone from two lines that occasionally touch to two perfect duplicates, one just offset from the other.

  5. Steve,

    I put the charts together myself. I can assure you that the only thing that is different between the two charts are the axes.

    Keep in mind that the first chart’s values are a lot closer together, which puts the lines closer together.

  6. agreed steve, 1/31/08 info shows that they are not the same. still point taken about lies and statistics.

  7. Jeff,

    I promise you, the same data is in both charts.

  8. Ok, finally got it, you’re using two different y-axis on the charts.

    It’s a good thing I’m more of a numbers man than a chart man, because I’ve never noticed that trick before.

  9. Another favorite is with bar charts. One graph shows entire bar chart with a 10% difference between two values and second version shows just the upper portion of the bar chart. 10% difference can then look like 50%+ difference with the same numbers.

  10. Bobby,

    That’s a very good point! Thanks for sharing.

  11. I love it when I see a graph with NO scale. Yea, okay.