This post is about the numbers around us. About the numbers, graphs and statistics people love throwing at us to prove their point.
How much do we really understand about these numbers?
My take – not very much!
A joke I once heard illustrates how “well” people understand statistics:
An American woman with 4 children says to her husband: “I just heard that one out of four children around the world is Chinese. We’ve been lucky so far”.
This is a classic example of someone who understands nothing about statistics. She’s got the numbers right but hasn’t got the faintest idea what they mean.
(explanation for those who are just as clueless as the woman in the joke above – although 1 in 4 children is Chinese they’re probably born in China and not in the USA).
But sometimes numbers are correct and it’s just a question of how they’re presented.
Here’s a classic example:
Actual case: A 100m race between an American and a Russian in which the American won.
Headlines in USA the next day: American beats Russian in a USA-Russia race.
Headline in Russia the next day: In an International race – Russian 2nd, American one before last.
The same case, using the same numbers but presented in a completely different way.
Why did I decide to write about this?
Because of an article I heard on the radio the other day talking about the percent of children killed in car accidents in Israel. They wanted to prove that children in Israel are at higher risk than children elsewhere.
Now don’t get me wrong – this is no laughing matter and regardless of whether 1 child or 100 were killed its tragic and should be prevented.
However I don’t like being brainwashed by mindless statistics.
You cannot say that Israeli children are more likely to get killed in car accidents just by comparing the percent of children out of the total number of casualties. You need to take into account the actual percent of children in the population in the different countries.
Unfortunately I can’t find the data from this year, so I’ll use data from previous years to illustrate. To make this shorter I’ll only include some of the countries in the articles.
Here’s some of the data from 2003 regarding percentage of children killed in car accidents**:
Turkey – 13%, Israel – 11%, Luxembourg – 9%, Estonia – 9%, Ireland – 8%, USA – 5%, Russia – 4%
Now let’s look at the percentage of children out of the total population (Statistical disclaimer – No 2003 data, used the data for 2004 as I’m assuming it didn’t change much between 2003-2004)***:
Turkey – 30%, Israel – 28%, Luxembourg – 19%, Estonia – 17%, Ireland – 21%, USA – 21%, Russia – 16%.
So now if we put these numbers together to understand what percentage of children are affected we get the following (divide the first number by the second to get a percentage, for Israel 11/28 etc.):
Turkey – 43.3%, Israel – 39.3%, Luxembourg – 47.4%, Estonia – 52.9%, Ireland – 38.1%, USA – 23.8%, Russia – 25%.
Based on these numbers – Estonian children are more likely to get hurt than children in Luxembourg, Turkey and Israel. Ireland also seems nearly as dangerous as Israel. Russia is more dangerous than the USA although the initial numbers suggested the reverse.
Here’s another hypothetical example (I made this one up):
Recent study shows: UK roads are more dangerous for English speakers than other EU countries.
Statistics to back this up: Data showing the percent of English speakers involved in car accidents in various countries:
- UK – 95%
- France – 15%
- Germany – 21%
- Italy – 18%
- Holland – 22%
- Scandinavia – 33%
- Eastern Europe – 7%
So what’s wrong with these numbers? Don’t they prove the point in the headline? Of course they do, if you don’t take into account that the UK is the ONLY English-speaking country in Europe and is therefore bound to have a much larger percentage of English speakers. This obviously affects the way these numbers should be perceived.
Why am I going on about this?
Because the media loves shoving numbers in our face to prove some point or other and most of us out there are duped because we lack information or knowledge.
Do we know that the sample used was large enough or that it wasn’t biased. I mean I could easily prove that people in the middle east have seen camels more often than people in Europe or vice-versa regarding squirrels. Does that tell us anything about people’s eyesight in different regions of the world? NO. Why not? Because camels don’t roam around freely in European parks and squirrels do.
I could also prove that people eat less sugar and carbohydrates than they did 2 years ago. How? I’ll just poll people who’ve discovered they’re diabetic in the past 2 years.
My point?
Beware of numbers! Statistics can prove a lot but they can prove to be very unreliable if they’re not done properly or if they’re marketed inappropriately.
Data sources:
** 2003 data of children killed in car accidents: Or Yarok Research
*** 2004 data regarding World population and percentage of children: Population Reference Bureau (PRB) Data