stats

Lies, damn lies and statistics – what can we really trust?

“There are lies, damn lies and statistics.”

Mark Twain is the alleged author given credit of saying this back in 1907.

And it is just as true today as it was back then.

I got first-hand experience of this mantra many years ago with this chart about the Polio eradication:

stats Polio vaccine misinformation

A glance at this chart proves the Polio vaccine wiped out Poliomyelitis, right?

But this data is carefully cropped.

We do not see that just 13 years earlier, Polio was naturally on the decline due to better hygiene:

stats Polio vaccine truth

Are statistics reliable?

I make heavy use of statistics on this blog.

I do so, because people gravitate to them like bees to honey.

We have been trained to accept statistics as truth.

This is especially true when delivered by prestigious people. You know… the ones who often wear white lab coats and hang framed, embossed papers on their office walls.

But the reality is statistics are almost always unreliable – most times they are flat-out lies.

Spotting statistical abuse

I use statistics the opposite way they are intended… to spot impossibilities and errors.

Statistics offer me a springboard to find ultimate truths.

Here are the 7 ways I am able to spot inaccurate statistics in a blink:

The sampling trick

Thank you John Adams for sending me this statistical tomfoolery:

stats why women feel cold

This article titled “Why women feel cold in the office” is a typical example of statistical shenanigans.

The study claims women working in an office prefer it about 5 degrees warmer than men.

Of course, we all know women typically wear less clothing in the office than men do…

But hey, why let the facts get in the way of a great statistics’ story, right?

And get this: the study’s sample size was just 16 women.

Let me repeat this. Just 16 people participated in the study!

This is like flipping a quarter twice… it lands on heads… and we assume flipping a third time will most certainly land on heads, too.

The statistics would be more reliable if the study surveyed 1,600 women – not 16.

Clearly, this study does not meet the minimum participants needed to be valid…

But that does not stop over 5,500 news outlets from reporting this.

Who benefits from the outcome?

I am always on the lookout for the “cui bono”…

That is, who benefits from these statistical outcomes?

In this same study, we read:

Energy consumption of residential buildings and offices adds up to about 30% of total carbon dioxide emissions; and occupant behavior contributes to 80% of the variation in energy consumption.

So, this is clearly another propaganda piece trying to convince us that we humans are the cause of climate change.

To confirm the mission of this statistical study, we see the heading “nature climate change” on top of the report.

(Repetition is crucial for biasing statistics.)

If this was not enough propaganda, this study pokes fun at men… and fat people.

A trifecta.

Sweet!

Who funds the statistical study?

Missing from this is who financially backed the study.

This is always a red flag for me.

Getting cash grants for biased statistical studies is quite common.

Because in the scientific community, revealing the truth might be a fast path to bankruptcy.

“Science” almost always points in the direction of the money train. And that train is often going in the wrong direction.

Magic stat graphing

There are subtle schemes statisticians use to warp reality.

Missing graph data, shortening time spans – even flat-out lies can turn a simple chart into a nefarious propaganda message.

Remember this ad from decades back:

stats Gerber lie

The ad claimed 4 out of 5 pediatricians (or 80%) recommended Gerber baby food products.

But according to a 1997 Federal Trade Commission report, just 16% of pediatricians recommended Gerber to their patients.

Oops.

Lies. Damn lies. And statistics.

Correlation does not imply causation

Even if the study data is perfect in every way, we cannot always jump to conclusions.

Let me prove it with a silly example:

Did you know that red-headed men with mustaches are more obese than the average man here in America?

This is a fact (according to me) :>

Because every overweight red-headed person I know is heavier than normal. And these two people both have mustaches.

But, the reality is they are probably overweight because they digest processed and refined foods.

In simple speak, there are many possible patterns that afflict red-headed, moustached fat guys.

I never forget that correlation does not imply causation.

The Hawthorne Effect

Researchers studied worker productivity for 8 straight years in the Hawthorne Works.

They tinkered with factory lighting, cleanliness – even moving workstations around.

But just as researches noticed improvement from the changes, productivity plummeted as soon as the researchers went home.

It was the worker’s knowledge of the experiment, not the changes that was responsible for the boost.

Public-reviewed findings (with open commenting)?

Imagine releasing statistics to the public for peer review?

This allows a worldwide selection of strangers to check the work free.

The comments would offer valuable insight – giving us the ability to do our own homework.

But the vast majority of statisticians cannot offer this, because their funding source would block it.

The punchline about statistics

Whether it is fair or not, I do not trust statistics.

I have more faith in my gut than some chart or graph put out by statisticians.

The good news is our intuition is almost always right.

Published by

Markus Allen

Family man. Truth seeker. Life hacker... more about me here...

 


Enjoy this post? Spread the word - share it:
share this articletweet this articlegoogle plusemail this to your friend

Feedback | Tell Markus you like this daily blog...

 

Click here to see my entire archive of daily blog posts...