In an excellent representation of Statistical Data by author “Darrell Huff” in his book “How to Lie with Statistics”, He has almost proved that how we are fooled in everyday life and almost all times by the use of statistics.

On reading this book I realised, and even you will realise on reading this summary that how each statements, data, averages, numbers, figures, percentages, charts, etc. are twisted to provide a glossier or worse picture then the reality. Or to even provide you an unrelated data so that you reach an unwarranted conclusions (which benefits the publisher of statistics in his own way).

This summary is especially for those people, who read lots of informative content and based on which make important decisions. Decisions like Investing, Buying, Selecting something, etc. People reading a lot of content published by Companies, Journalist, Writers, Authors, etc. across Annual Reports, Newspapers, Blogs, Articles, and Books and so on will realise how small things are made up to appear in such a way so as to mislead you and your decisions.

Also, there are many tips in the book to tackle these misrepresented statistics and same has been summarised in this post.

So, Let us now go through the quick shortened summary of a small short book by Darrell Huff to open our eyes to Fallacy.

• Misleading 1: The Twisted Statements

Many a times you will find contents where the sentences have been twisted in such a way that you will only understand what is easy and ignore the tougher part. Also, when a stat is presented, it is not always related to the conclusion drawn. Those stats are only presented to mislead you and create an illusion in your mind in accepting something.

 “Proper treatment will cure a cold in seven days, but left to itself a cold will hang on for a week.” – Imagine this is published by a company producing tablets for Cold. For once, break the sentence into two halves and you will realise they mean the same. Well, on general reading it implies that to cure a cold in seven days you should take medicines. These sentences, when written separately are understood better and in meaningful way, but imagine all those small sentences hidden in the huge article that you read. You will definitely move on, assuming this medicine is good enough to treat your cold in only span of seven days.

There are many such sentences, often twisted or placed in such a way that your conclusions are altered. “If you drive car at 70 miles/hour on a highway, you are safe four times better at 7am then at 7pm”. You decide. When is the peak traffic? At morning 7am or in the evening. So, when you are safer? In fact, a study has also stated that, when you drive car at 7am, you have more chances of accident due to visibility issue because of fog. But, this sentence is so nicely framed to encourage you to drive faster in the morning and hiding the real part of the stat.

“Hence, when a stat, directly is not favourable to your side, advertisers compare with other stat or other picture to make it look favourable.” Say for example, an automobile company. They state, railway saw a surge of 100% accidents as compared to last five years. Well, that might seem horrific and you may avoid using railway at all. But, the number of passengers travelling in railways have surged 20 times and that is the part hidden. Now, this stat is only presented to you, so that your decision to use railway is altered and preference towards using automobile increases. But, what they do not present is that automobile accidents have increased 3 times as compared to last five years.

• Misleading 2: Sample Error and Survey Conclusion

Have you ever come across with many of these similar statements that states “A Study conducted by well-known dentist, he has arrived at conclusion that people using “This Toothpaste” have improvement in their tooth decay.” You will find many such studies. But, the question is: study was conducted on what type of people and how many of them? It was dentist conducting the survey only for those people who were his patient? If that is so, obviously when patient come to doctor to treat themselves, they will definitely find improvement in their decay. And this improvement was used by companies to state that it is only because of the toothpaste they made.

Flip a coin 10 times, you have chances of getting 7 heads. But flip a coin 10000 times and the ratio of heads to tails will almost be similar. This is another thing with which the stats are worked out. There are many companies conducting survey on only limited number of samples. They repeatedly take the surveys year on year, once they find in any year that the result of survey are favouring them, soon you will find a statement is bold and capital letters “This year, there has been a superb shift in consumers using Herbal Medicine.” What we fail to realise is that: One, the stat has just luckily played out due to very small sample size and second, the category of people chosen might be old aged so that the stat works out favourable for company.

Hence, two things to focus on. The Sample size through which the result of survey is taken out, and the suitability of sample taken to represent the whole audience.

• Misleading 3: The Number Game.

This is the most favourite statistics of the publisher. They will show you some exciting number to state that we have conducted research ourselves and the results are so astonishing to believe.

Take e-commerce companies for example. “Our sales have increased a whopping 1500% as compared to past 7 years.” This is so good for a statistics. But, have you ever imagined why they compared the past 7 year’s figure. Why not last years? And Why not last 5 years? Well, there is reason for everything. It might be possible that sales have declined when compared to last year, and for last 5 years, the sales have only rose 100%. Both numbers for past year and past 5 years would look ordinary, but compare it with past 7 years, and anyone will say, Wow!, your company is doing great!. This is called BASE RATE for comparison. BASE chosen here is the past 7 years. The base is played out/chosen to alter the output. Like, a company have Net Profit Margin (Profit/Sales) of 1%, but they might be earning 25% on the amount they have invested. To show a worse picture: Your company has only earned marginal Net Profit of 1%, and a better picture: This year, we have achieved a whopping margin of 25% on our investments. The BASE chosen here is Net Profit and Investment and by changing the base the output is altered drastically and so is the conclusion.

Percentages in stat is also altered. A 1% v/s 25% difference is huge and the one who might not understand the crux, will find this very interesting. Percentages, Numbers, Averages, etc. everything are played out to alter the conclusions at the readers end. “There are often many ways of expressing any figure. You can, for instance, express exactly the same fact by calling it a one per cent return on sales, a twenty five per cent return on investment, a ten-million-dollar profit, an increase in profits of forty per cent (compared with 2015-16), Or a decrease of sixty per cent from last year. The method is to choose the one that sounds best for the Purpose at hand and trust that few who read it will recognize how imperfectly it reflects the situation.”

And the same perfectly goes when presenting an average. Averages are Simple, Weighted, Geometric and further into Mean, Median, Modes and so on. Smartly pick up only those average which will favour your report. Also, add a pinch of sweetness by altering the sample size and you have set a very strong stat to present. “Average monthly salary for people residing in Mumbai is Rs. 36,600/-, this show’s standard of living amongst people are improving.” Let me tell you, Mumbai is financial capital of India, and there is huge difference in earnings of different people across this city and across country. You will find extremely rich people in Mumbai and the average income will shoot up like anything. And when this is played with Mean, Median and Mode, you can chose the type of average you want for the number and statistics to appear awesome.

But, again How can one say the pin point number of Rs. 36,600/-? Is it that, the surveyor has reached all the people residing in Mumbai and found out these stat? No Way, right? One cannot come up with a pin point number definitely. So, when the numbers are presented so accurately you have a doubt in the statistic. Ranges are hence more preferable. When it is stated that “However, salaries range from Rs. 9,260/- to Rs. 1, 63,000/- depending upon the earning source of people.” It looks more reliable underlying the realty.

So, next time you come across reading these types of stat. Be Careful!!!

• Misleading 4: The Visual Presentation – Charts, etc.

This is somewhat related to Number Game and choosing Base. Charts are used to show some trend in numbers. Mainly Line and Bar charts are used. But, even the charts are presented very smartly. The x and the y axes of the charts are altered in such a way that the lines/bars appearing in the chart seems very promising. People tend to ignore the axes, numbers, etc. and focus more on the lines, bars and its visual trend.

Say, there is company whose profit are continuously decreasing year on year. I have a trick, instead of showing 2015, 2016, 2017 in your y-axis, show it reverse i.e. start from 2017, 2015 and then 2016. The line chart from left to right will show an increasing trend even though the profit has decreased. This is done just by altering the y-axis and the same can be done with x-axis as well. When your scale in x-axis is 5-10-15-20, change it to 2-4-6-8-10-12-14-16-18-20. The chart will be altered and a rise from 5 to 15 will appear huge in changed axis of 2 points when compared to axis of 5 points.

Similarly, there are many other visual representation through which a stat is exaggerated. Have a look:

• Mislead?

So, I just have explained you and tried to cover all the misleading stat which can often lead to unwarranted conclusions and decisions. There are lots of ways to alter the statistics. Often, publishers find new ways to present the stat. So, it does not mean you will not be able to make out the real stats v/s the made up stat. There are many ways by which you can find out.

“Give that kind of second look to the things you read, and you can avoid learning a whole lot of things that are not so.” And “No statistical conclusion should be read without having sceptical question in mind that the stat might contain biases i.e. to find out how less biases or how much more biases are reflecting in the presented stat.” There are the lines picked up directly from the book.

Author Darrell Huff has also presented ways through which you can find how real the statistics are and does it contain any biases. Here are the 5 point instruction before you believe in any stat.

• Understanding Real Statistics – Instructions.

So, I just have explained you and tried to cover all the misleading stat which can lead to bad decisions and conclusions. And there are also ways by which you can avoid them. The most important is that you should learn to question the presented statistics. Author have laid down a 5 question instruction, which if you find answer to, be able to judge the difference between a true stat v/s made-up stat. Those questions are:

1. Who Says So?
2. How Does He Know?
3. What’s missing?
4. Did somebody change the subject?
5. Does it make sense?

Ask these question and you will get answers to the reliability of a statistics. Whenever a stat is presented, always see who has conducted the survey/research. Just as a doctor is needed to comment on particular disease, you need an Independent expert of that particular field to survey on the statistics. This will only give you a stat on which you can rely at first instance. But, what if that independent expert is not really independent. How can you verify that? Try to find the answer to “How does he know?” and your problem will be solved. He know by conduction a research on all his patients. You will conclude that the expert is biased towards his scope of study and cannot be stated as independent at all. And this will also open your eyes to ask next question “What’s missing?” and obviously the sample size. I.e. only a handful of doctors patients are only used to arrive at the conclusion.

Hence, the first three question are the chain that comes one by one. And also there are certain facts/statistics which are absolutely correct, but completely irrelevant. Just like accidents in Railways are linked to Automobiles. Hence, although correct statistics of railways are presented, they do not matter at all for the subject/article being referred to. Hence, the last two question will help in catching that misrepresentation.

“Did somebody change the subject?” will help you recognise any irrelevant statistic that is being presented to make the other part appear good. Just like automobile accidents v/s railways accidents. And also “Does it makes sense?” helps you to identify the same. Any glossy stat is presented in middle of any paraphrase, it does not mean it is very relevant here. Actually, it does not makes sense at all to show this stat in midst of some unrelated paragraph.

• Closure

And so, this comes to the end of summary of the Book “How to Lie with Statistics” by “Darrell Huff.”

I hope you like my short summary of what the author has to say regarding “How to lie with statistics.” Well, the writers might influence you but the readers should be smart enough to not fall into the prey. Thank You!

 

Disclaimer: I have tried my best to evaluate the content from book into a short summary, and where necessary added my own point of view. By no means, this summary is an infringement to any involved party.