Deception with Statistics

Techniques to Deceive with Statistics

Statistics is the science that deals with the collecting, classifying, analysing and interpreting numerical information. The findings from all that activity are also called statistics. So, statistics (the science) creates statistics (the results). Both the data-gathering process and interpreting the data can be manipulated to present the answer you want.

Here are some common techniques used to manipulate statistics: Statistics are the lifeblood of private companies, politicians and governments. They all use statistics to influence you and to manipulate your actions. For example, they will use them to persuade you to make investment decisions, to encourage you to buy their products, or to show you they are tough on crime or skilled at managing the economy.

However, manipulative companies and individuals have become highly adept at spinning statistics to make the "facts" look more favourable to them.

You Can't Trust the Experts

Never forget that lots of the statistics you see are not designed to inform you but to influence you to buy something or to adopt someone else's point of view. And also bear in mind that some of the worst offenders are the respectably titled experts who operate in fields you know nothing about. These people are actively spinning statistics to influence you. It's their job. So, we all need to be far more sceptical about statistics. (Even if the statistics are not outright lies or obviously biased, there's a fair chance they're being presented by people ignorant of or indifferent to their accuracy.) Also, remain mindful that when you use statistics (and often your audience will expect objective empirical evidence to back your arguments), you expose your reasoning to attack. Don't be afraid to use statistics, just make sure they've been properly researched and are intellectually defensible. The challenge is to have factored in all the variables before you present your statistics. It's quite hard for someone to check your underlying data. It's far easier for them to find a confounding variable (i.e., a factor you've not considered). And don't forget to examine the reverse logic and be prepared to explain it. Also, use as many samples as you can to lower the P-value, and make sure they're not from a biased pool.

Scrutinise the Language

Let's look at the words of a motor-insurance advert that I saw recently. The ad featured a car mangled from a crash and a claim that car crashes cause 2,538 deaths per year. It also included: "1 in 200 people are killed in car crashes". I thought that the "200" figure looked a bit low, so I went trawling for some statistics to back it up. I couldn't find any. So, did they just make up that "1 in 200 people are killed in car crashes"? Possibly. But, in my experience, these people are usually a little bit more conscientious than just inventing figures. So, were they spinning the truth with their language? Probably, and this is my suspicion.

Let's look at those words again "1 in 200 people are killed in car crashes." What does that mean? Does it mean the cause of death for 1 in 200 of us will be a car crash? Well, it could mean that. But, it could also mean that 1 in 200 of us will die if we are in a car crash. The English is ambiguous.

The use of ambiguous language is just one technique for influencing you. Look at the language used in these claims:

• "Scientists have proven it will make your skin smoother."
(Whose scientists? This is an example of an appeal to an unnamed authority.)
• "It contains polypeptides that help the cells communicate."
(This is an example of obfuscation fallacy.)
• "Formulated with active Dead Sea minerals, our vitamin E-infused nourishing cream is enriched with ivory-coloured Shea butter and organic jojoba oil for continuous hydration."
(Coupled with a high price tag this is an example of the creating the expectation effect (i.e., when something is considered better because you expect it to be better).)

See Also