The Power of Words: Use & Misuse of Statistics

This is the first article I am copying over from ILPPS. It’s an important lesson about media and statistics.

“There are three kinds of lies: lies, damned lies, and statistics.” – Benjamin Disraeli

Please read the following two hypothetical statements and answer which of the scenarios shows an increase of 20%.

wiki

“In clinical trials, patients who ate twice as much red meat saw an increased heart disease rate from 5 per 1000 to 6 per 1000”

“In clinical trials, patients who ate twice as much red meat saw an increased heart disease rate from 5 per 100 to 25 per 100”

The answer of course, as anybody who knows even the most basic concepts of statistics, is both. Yet when you look at the actual figures they are vastly different. An increase from 5 to 6 per 1000 is so negligible, it’s not worth getting excited about. An increase from 5 to 25 per 100 is vast, statistically significant and of major concern. So how can they both be an increase of 20%?

The answer is in how the figures are presented. The increase from 5/1000 to 6/1000 is a 20% relative increase and the increase from 5/100 to 25/100 is an absolute increase. Now think about the sort of headlines we see every day. Knowing this information now, and realising that the media and other organisation leave out this vital piece of information, I hope you will look at the sort of statistics that get thrown around in a completely new light.

Look at this article from The Daily Mail “A woman’s risk of breast cancer increases by 6% for every extra alcoholic drink she has on a daily basis, researchers said today.”

Then this one from The Guardian. “The IARC’s experts concluded that each 50-gram (1.8-ounce) portion of processed meat eaten daily increased the risk of colorectal cancer by 18%”

I could go on and find other “X increases chances of Y by Z%” but I’m sure you get the idea already. All of them will point out an increased risk of getting something from doing something, but whether it is relative or actual is never discussed.

  • Relative increase – A trend relating to variations from the original data (from 5/100 to 6/100, from 5/10,000 to 6/10,000 or 5/1,000,000 to 6/1,000,000 is always a relative increase of 20%)
  • Absolute increase – An increase in actual percentage (from 5% to 25% is an absolute increase of 20% as is a rise from 30% to 50%, from 50% to 70% etc)

This is a vital distinction to make; I can’t state this enough, especially if we are talking about food stuffs causing specific diseases such as cancer as happened in 2015 (and several years ago when discussing the carcinogenic nature of bacon) which results in people worrying unnecessarily.

Groups such as The Vegetarian Society and the Organic industry love to point out that eating red meat/any meat will increase your risk of certain diseases, especially cancer and they will never explain whether the increase is absolute or relative. I prefer to give people the benefit of the doubt and assume that writers who compile these stories don’t know the difference, and that’s why it is important that we educate ourselves to challenge and look a little deeper at what we are being told by the media.

Depending on how the figures are being presented, the same percentage increase could be either negligible and therefore not worth worrying about, or significant and cause for major concern. Next time you are presented with the sort of stories we see above, ask yourself what it is they are actually saying.

Advertisements

3 thoughts on “The Power of Words: Use & Misuse of Statistics

  1. But I still don’t know whether I should avoid bacon or not!

    Great post, but honestly, I still don’t understand what the Hell a “relative increase is”… I’m clearly not a numbers person…..

    That picture of the bacon though…

    1. Basically, a relative increase is the increase on the existing number. An absolute increase is an increase in real terms and nothing to do with what the pre increase figure might be. If a high number, the latter is significant and the former not necessarily so.

      Let me put it another way.

      If your chances of getting cancer goes up from 3% to 4% by eating bacon, you would think that that was so insignificant as to not even worry about it. If your chances went up from 3% to 36%, you would probably consider giving up bacon.

      Yet both numbers could be considered a 33% increase and the former sounds worse when you say “Your chances of getting cancer are 33% higher if you eat bacon”. 1 is 33% of 3. Add that to 3 and you have 4. Add 33 to 3 and you have 36. The former we call relative and the latter is absolute.

      Is that clearer?

      1. Thanks so much for taking the time to explain it to me! But I’m seriously hopeless. I think my brain just sees numbers and percentages and starts maniacally repeating the mantra “cannot compute”.

        It’s okay though. I’ve made peace with it 🙂

Have something to say? Go on, you know you want to:

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

WordPress.com Logo

You are commenting using your WordPress.com 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 )

Google+ photo

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

Connecting to %s