Finished reading this incredible book ‘How to Lie with Statistics’ by Darrel Huff. It’s been a while since I read a book – hectic work schedules at office coupled with preparations for joining full time MBA program this summer have kept me really busy. The book itself is a part of the reading list given to me as a part of pre-MBA coursework.

The book, as the name suggests, talks about how statistics can be (and are) used as tool of deception – for distorting facts in a way that is suitable for an interested party. The book has been sectioned into 10 chapters most of which introduce one or more ways in which statistics are used to give the illusion of reality that is different from what really exists. The writeup is full of examples – from magazines, newspapers, public reports and surveys- to illustrate the use of these tools and to help the reader understand what to look out for when he/she comes across an impressive statistic. The last chapter of the book is dedicated entirely to helping the reader understand how to critically evaluate a beautiful picture painted with the help of statistics.

Here’s a look at what all the book talks about:

**The Sample with the Built-in Bias**

Would you believe a doctor if he said that most of the people in the world are ill because of all the people he has met in the last 1 year most were? You wouldn’t, right? Being a doctor he obviously meets more people who’re not keeping well than those who are perfectly healthy. However, most people do tend to believe a lot of statistics before trying to verify if the sample of population, on which the statistic is based, accurately represents the said population. The writer provides multiple examples of how biased samples lead to inaccurate results. On occasions the bias is introduced to obtain results that suit an involved party.

**The Well·Chosen Average**

You leaned about mean, median, and mode during school, didn’t you? Then why is it that when you read the work Average in a report or ad you don’t stop to ask which of those three are being talked about? Wait to think for a second- when discussing salaries what are you more interested in- median (*the figure that accurately tells you half the people take home more money than the said amount and the remaining half take home less salary than the said amount*) or the mean (*a figure that is hardly representative of the state of salaries among the said group as it can easily be brought up or down by extremely large or extremely low salaries that a few might be drawing*).

**The Little Figures That Are Not There**

Ever wondered how a toothpaste brand can claim that using their product helps reduce the chances of tooth decay by so-and-so per cent without getting sued? They do it. It has to mean that the results are genuine. This chapter of the book deals with such mysteries- how interested parties at times withhold important information to make the statistics seem different from what they really are.

**Much Ado about Practically Nothing**

A difference is a difference if it makes a difference. This is what the author explores in this chapter. He introduces the concept of errors and ranges to show that at times difference between numbers are made to mean more than they do.

**The Gee-Whiz Graph**

Statistics are mere number and most people are bound to find them boring. However, if you plot these numbers against the two axis of a graph the result is a interesting diagram. Now, how does one make this interesting diagram even more appealing? If only there were a way to make the red line climb (or drop) much more sharply. Aha! Now that’s an idea!

**The One-Dimensional Picture**

So graphs can help statisticians attract the attention of the general folk towards (only) the aspects they want to. How do you dramatize further? Pictures! This isn’t misleading you say. Well, consider this- to show that the per capita income of people in a Region A is twice as much as that of people in Region B, I draw 2 pictures of money-bags labelled A and B with the bag twice as tall as the bag B. Interestingly, to make the bag A twice as tall while making it look somewhat similar in shape to bag A I make it twice as wide as well. Most people who look at the diagram would imagine that the bag A is twice as deep as bag B. What all this means is that the volume of bag B is 8 times that of bag A (it can hold 8 times as much money as bag A). :-) Simple trick to fool my readers into believing something without even saying it.

**The Semiattached Figure**

If you can’t prove something then demonstrate something else and pretend they’re the same thing. That’s how the author begins this chapter, and frankly that’s all that he talks about. The examples with amaze you! Here’s one of the more basics ones-* More people were killed by airplanes last year than in 1910. Hence, modern planes are more dangerous*. Ridiculous logic you’d say. The statistic might be in place but there’s a big assumption that separates the statistic and the conclusion (assumption- the same number of planes fly in the skies today as in 1910). It was easy to spot in this case, but is it always this easy? ;-)

**Post Hoc Rides Again**

A friend once told me that he had read somewhere that high heels are popular among ladies when the stock market’s doing well and flat shoes are more popular when the stock market’s not doing well. Interesting fact; however, could one conclude that if one somehow found a way to make high heels popular among ladies, he/she could make the stock markets perform well? The author explores similar correlations in this chapter and brings out four kinds of fallacies- correlation by chance, correlation where it’s impossible to detect whether A is causing B or otherwise, correlation where neither A nor B is causing the other (both are being caused by C), and the correlation that’s inferred to hold true beyond the data with which it has been demonstrated.

**How to Statisticulate**

In this chapter the author introduce Statisticulaton – Statistical Manipulation. He talks of deception through the decimal, through percentages, by adding things that don’t really add up, through percentiles, and through averages.

**How to Talk Back to a Statistic**

In this final chapter of the book the author shifts focus from revealing statistical deceptions to revealing the secrets to how statistical deceptions can be revealed. He shares 5 questions that can act as effective devices for critically examining statistics:

- Who says so?
- How does he know?
- What’s missing?
- Did someone change the subject?
- Does it make sense?

Read to find out how you can make use of these questions to talk back to a statistic.

ridhima jain

June 27, 2011 at 1:18 PM

Nice article. Really inspired me to pick up the book :)

Christie

August 23, 2012 at 9:13 PM

I read this book for my AP Stat class this summer. Its very insightful and the author says everything in a very easy to understand manner. I really enjoyed it, and I think everyone should read this book!