Ex-lawyer turned relationship coach

Statistics End Arguments And Start Trouble

Statistics are persuasive. The most important decisions in the world are often based on statistics. But statistics are also misleading.

Back in 2015, ex-congressman Jason Caffetz was trying to shut an organisation about women’s health and flashed the room a chart to get his way. The chart showed two lines crossing, one soaring up in pink (abortions) and the other slumping down in red (cancer screening and prevention services).

Caffetz argued, “Why should we fund an organisation that isn’t saving lives?”

The problem?

The crossed lines suggest a conclusion that’s wrong. Looking at the chart, you get the idea abortions are going through the roof. But if you check the numbers, the line that represents saved lives is three times greater than the line that represents abortions.

The chart was purposely misleading.

As is the following headline from the New York Times, “1 in 4 Women Experience Sex Assault on Campus.” 

An incredibly shocking number. That’s greatly inflated.

What’s the issue?

Only two universities participated in the survey that introduced the statistic. More than eighty percent of female students didn’t participate in the survey. And the poll takers never defined ‘sexual assault’. 

In fact, the survey didn’t even use the word ‘assault’. So ‘sexual assault’ could mean anything from an uninvited kiss at a party to a brute violation at gunpoint.

While both scenarios are bad, one is clearly worse than the other.

Just like you’d prefer a lifetime with fewer economic recessions than more. Which actually has less to do with the economy and more with statistics.

Take a look at the United States, the US just experienced its fastest three months of economic growth on record. But the economy is still in a hole.

The states simply rebounded back to normal after a terrible second quarter. Is the recession over or still going?

It depends on who you ask and when you crunch the numbers.

Which brings me to a statistic that’s so misleading it earned itself a name, Simpson’s Paradox. A trend that appears in different groups of data but disappears when these groups are combined.

Confused? Here’s an example, “Smokers have a higher survival rate than non-smokers.”

Sounds like bogus, but that statement is true when you compare the right groups. Like when you measure the mortality rates of teen smokers with the elderly. However you slice it, the elderly are most likely to croak first.

The Simpson’s Paradox and its other misleading statistical siblings show that numbers don’t have to lie to trick you. So next time you’re fed a graph or percentage, see it for what it is: a collection of numerical data organised by an imperfect human.

By Jeroen Elsing
Ex-lawyer turned relationship coach