How to lie with Statistics

This has been one of my favorite reads of all time. At first I bought it solely as a book to get a laugh from people with, but as I began to read it I discovered there were lots of ways to lie with statistics that I hadn’t thought of myself yet. As always, I am not a proponent of activities such as this, but I feel it is very valuable to be aware…

Here are a few of the ways he gives to lie with statistics…

Use images instead of graphs.

Lets say you want to illustrate that salleries have doubled in 10 years. But you want it to make an impact… So you prepare an image that has a dollar sign over a label that says “Then” and then you have a dollar sign that you have scaled by 2x with the label “Now”. This is very misleading, because when you scaled the money bag by 2x, you didn’t just make it 2x as tall, you made it 2x as wide… Effectively making it 4x it’s original size! This makes for a MUCH better impact than just a simple bar graph.

When presenting a graph make the scale very small or don’t include one at all.

You can make almost any “growth” look phenomenal if you pick the right scale. Lets say your sales grew $100 dollars. This is huge if you are a ten year old running a lemonade stand, but insignificant if your a fortune 500 company. So, draw your graph like you are a lemonade stand… Make the bottom of the graph be last years sales, and make the top of the graph last years sales + $100. It’s easy! Be careful with this one though, because if you go too far, it will be obvious.

Never play when the odds are fair.

One of my favorite Star Wars characters once said something along the lines of “Never play if the odds are fair.” (He said this prior to fixing an election on a small planet.) The same is true with marketing. You never ever want to commission a survey with the intent of publishing the results without first doing something to ensure your own success.

There are lots of ways of commissioning this study with the odds in your favor.

1. First, send salesmen to 100 dentists and give them hundreds of samples of your toothpaste. Explain to them all the benefits of your tooth patse, and then ask them if they would recommend your toothpaste to clients.
2. Ask your sales dept for a list of all the dentists that have ordered your toothpaste and send them surveys asking them what toothpaste they recomment.
3. Hire four dentists on your staff, and after 3 months ask them which toothpaste they would recommend.
4. Send lots of your toothpaste to a small third world country. Then, send surveys (with promises of more toothpaste for each returned survey) to each of the dentists that received the paste.

Pick the average that suites you best

What most of us (unless your a marketer) have forgotten from high school math is that there are 3 different ways to get an average… (Watch the video.)

Mean Median and Mode

As that video illustrates… When you say “Average” you actually get to take your pick from one of three numbers (mean, median, and mode)… And you aren’t required to say which! So, just pick the one that looks best, and stick with it.

What does this mean for me?

Dig behind the numbers. Never take a statement from an untrusted source at face value. (and never trust an advertisement) If it’s true, there will be evidence to support it. In the words of a very very very smart person… “By their fruits, ye shall know them.”