Why Statistics Lie

Let me clarify, statistics don't lie.  They are inherently misleading.  The reason... context.

Correlation and Causation

In my AP Statistics class in high school, I was taught the golden rule of statistics.  

Correlation =/= Causation

This meant that the study had to come up with a hypothesis.  You would have multiple groups or trials conducted to test against that hypothesis, everything else being equal.  Run the statistics, and only if the groups exceeded a certain threshold (1, 5, or 10%) then you could conclude your hypothesis MAY be correct or you don't have enough evidence to say that your hypothesis is INCORRECT.  The test would have to be replicated over and over to get the same result.

Why the MAY

What do you mean MAY be correct or that it isn't INCORRECT?

It seems odd though a 99.999999% correlation my data seems pretty solid to just about everyone.  Moreover, it could get peer-reviewed and other people get the same exact result, but it still ends up as a MAY be correct.

This means that anyone could one day destroy everything we thought we knew about science with an experiment or fact that proves opposite. You always find that creationist use this in a debate against scientists.  If you could indeed find evidence that sediments could form in the 6000 years as stated in the Bible, then you could call into question just about any Atheist.  The problem is the evidence that they used to support a creationist view does not stand well against the scientific experiments that we have grown to accept as fact.

However, a common example that gets thrown out to demonstrate the fallacy is ice cream and drownings.  The amount of drownings increased as ice cream sales increased, so it seemed like ice cream caused drownings.  The problem was ice cream sales were increasing in the summer, and drownings were happening in the summer, because people were swimming in pools.  Ergo, correlation did not equal causation.

Women's wages

This is a pretty good example of where statistics are often misleading us.  In the fight for women's equality, we often see the quote for women making 77 cents for every dollar that men make.  This seems like a pretty big issue.  However, this is completely false.  If I was a smart business man, then I would just hire only women at my company.  I would save 20% of my wage expense, and I could out compete every male dominate company in the same industry.

What is the statistic?  It turns out it compares women median salary versus male median salary.  However, this gives no indication of what women's jobs are or men's jobs.  If we looked at that, we find that women are gravitating towards fields where that are lower paying, such as teaching, more than men.  Statistics show that men and women working the same jobs have a 4-7% difference in their paychecks, but it was unclear as to whether this was due to personal decisions or other factors.

The solution to this problem is not a national pay raise of 20% on all women's pay checks, but a closer examination as to why women are gravitating towards fields with lower salaries.

Democrats in office = better economy

This may or may not be true.  With a large issue such as the economy, it seems odd for a single party to claim that they are better at solving it.  However, this is what Hilary Clinton stated in hopes that the naive viewer would think that she is a better fit for office.  Although that may be what the statistic implies, we have to remember correlation does not equal causation.

Democrats may be better at solving the economy, or maybe they aren't.  It is hard to say, since we can't have a definitive case study with the way our country is structured.  For any 4-8 year President to claim the success of the economy is already pretty bold, especially since past policy also has an effect.

Conclusion

There are many examples of the fallacy, especially in the current media.  Hopefully, you are a little less rash when making decisions.

Peter