How To Lie With Statistics | Darrell Huff | Notes and Summary

This is an age where you find graphs and statistics thrown at you and in many cases tweaked in ways that lead you to believe things that are not right or are misleading. I read this book a few months back, but I felt it of prime importance to shed some light on this topic with the huge amount of misinformation and skewed charts that hit our screens leading us into believing narratives that may not be right for us. This book is much more relevant to us now with the Pandemic and the high volume of Statistics and Graphs that are being circulated around the social media. It is all the more important to make sure we see through the statistics and understand them.

The book How to lie with Statistics gives the common man insiders sneakpeaks of the the tricks statisticians hold up their sleeves. This is a fantastic introduction to shielding ourselves from misleading statistics.

Title: How To Lie With Statistics
Author: Darrell Huff
Pages: 142
Links: Goodreads | Amazon
Image Credits: Goodreads

Check the Samples of the survey/study

The results of a study are just as strong as the sample. Beware of samples that are biased or perhaps too small to talk about the entire population.

Most parents and students can perhaps relate to the advertisements where Universities and colleges talk about the average pay of all of their passouts. On face value it does make sense.

But think about how the college received the information. Did they take the average out of only the students who were placed from the college? If so, what about the other set of people who weren’t? For older batches, did they reach out to the alumini? How did they reach out? Did they reach out to them by mail, phone calls or door to door surveys? How many of the alumini responded to the mails? How open would they have been to revealing their income?

When statistics are based on opinions of people, there are high possibilities of the samples to be biased.

Trap of the Averages

Ever heard of people talking average household incomes? Or any average as a matter of fact. Keep in mind that there are three kinds of averages which indicate different aspects of the sample. Taking the example of average household incomes let me explain the three averages.

Mean gives the idea of the central value of the entire data set i.e. the sum of all the incomes divided by the number of households

Median gives the information of the income which lies exactly in the middle when all the samples are sorted, ascending or descending i.e the number of households earning more than the median income would be equal to the number of households that earn lesser than the median income.

Mode talks about the income which most of the households earn.

In case of household incomes, the median is generally the average that gives the actual idea about the sample. Always keep an eye out to check which average was used to derive the results claimed by the statistician.

Perseverance Pays

Have you ever felt something fishy when 9 out of ten dentists recommend a certain toothpaste? It is, in many cases! Give it a thought, what would be the odds of such high recommendations when almost all the toothpastes are made similarly and barely any difference?

This happens when companies discard data on innumerable experiments and wait until they get the results they want.

Hidden Errors

When talking about statistics, there is a high probability of error within the statistical process inherently. If you’ve seen advertisements about training minds and improving IQs, one must keep in mind that the IQ tested is just a fraction of the persons intelligence, most of these tests do not consider many forms of intelligence.

Also, many studies display results with standard errors e.g. two children with IQs where A has 98+/-3 and B has 100+/-3. At the first glance it would seem that B has a higher IQ, but if you look closely, there is a chance of A having an IQ of 101 while B has 97 as well!

In short, be careful whenever you find such numbers with errors.

Correlation Vs Causation

There are studies which say that students who score above average smoke lesser than those with below average grades. Even if the study was conducted carefully and meticulously, people jump to conclusions!

One can jump to a conclusion saying Smoking causes low grades. But there could be a possibility that scoring less pushed students to take up smoking. Or perhaps smoking is an effect of a third parameter altogether.

In such cases causality cannot be established, but a correlation is strongly visible.

Be careful before jumping to conclusions with the data set.

Manipulating the Graphs

It is also common that people manipulate graphs to prove their point! Check the graph below

The Democrats seem to be much much more compliant with the court right?

This is the right graph which gives a better understanding of the scenario. Now, if you look at it, the Democrats’ don’t look as impressive, right?
If you still aren’t convinced, perhaps the next picture would shed a little more light.

It is always better to look at the axes of the graphs to check if there has been any manipulations.

How to Talk to a statistician

Keep five questions in mind each time you come across any statistics

  1. Who Says so? – The source, do they have a point to prove? Are they affiliated to someone? Or an independent study?
  2. What’s missing? Talking in percentages always leads one to misread the statistic or even not mentioning the type of study or the average (mean, median or mode)
  3. How does the statistician know? Is the sample biased? What is the source of the study?
  4. Did Somebody change the subject? “More reported cases of a disease are not always the same thing as more cases of the disease” This is quite relatable in the current pandemic where the number of tests conducted also must be conducted. Low tests conducted give lower reported cases which doesn’t mean that the total cases are low. The remaining may not even have been tested!
  5. Does it make sense? Many of the statistics make sense when they are not looked at keenly but reading between the lines would give a better understanding of it.

The above notes are those that resonated most to me. The book definitely has many more concepts worth giving a read to help yourself and not succumb to tactics and many many errors you can be prone to believing in!

Also, do read 5 Ways Writers Use Misleading Graphs To Manipulate you. This article gives a clear understanding about misleading graphs.

Before you leave, perhaps you could listen to this…

I love the the transition between 3:15 and 4:30 !!

I wrote only about a few tricks about a statisticians sleeve and would recommend you to read the book. This 142 page book is worth a one day read with knowledge that would protect you from misinformation all through your life, and perhaps others as well!

Image Credits: 5 Ways Writers Use Misleading Graphs.
Featured Image Credit: Herding Cats.

8 thoughts on “How To Lie With Statistics | Darrell Huff | Notes and Summary

  1. Kushal, a very detailed, comprehensive and insightful post on the workings of the world of statistical data and its manipulation.

    Liked by 1 person

      1. I am currently reading ‘How to stop worrying and start living’. Very nice. Anecdotal and provides very feasible solutions.

        Liked by 1 person

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s