The Numbers Never Lie.

The numbers never lie is a line from a film I managed to watch on my flight from Seattle to Dulles Airport in Washington D.C.  Picture two guys dressed in homemade space suits sitting in a car, with the windows rolled up.  There are three rocket engines attached to the two sides and top.  The car is sitting on the same type of huge airplane that took the space shuttle up.  One of the guys is concerned it won't work and the other guy holds up his cell phone stating "The numbers never lie".  In reality, this situation would never happen since it is physically impossible.

In reality, numbers can lie to us or should I say, the presentation of numbers can lead us to believe things that are wrong.  This becomes especially important when looking at articles that present certain "facts". I watched a program one time where the doctor stated that coconut oil is good for us but I did some searching and discovered there is only one study out there supporting this statement and it only looked at 30 people.  This study had such a small sample size that the conclusion was not valid and that is one of the ways numbers can lie.  

A small sample size might be the number of people studied but it might also be sports writers coming to a conclusion about a players ability after a couple of weeks of play.  So it is always a good idea to see how many people were involved in the study, or how long a period of time the data is for when looking at a sports person. 

Another way numbers lie is when large meaningless numbers are thrown around. A good example of this is when looking at the number of followers an "influencer" has. The number may not reflect the actual number because there are many ways to increase the numbers of followers without caring if they all care.  I've personally seen where social myth busters have gone through and shown how one influencer increased their number of followers with entities who may have been created by her mother.

The third situation in which numbers may be lying involve correlation rather than causation. This is when one thing is said to cause something else when they are actually just a correlation.  An example would be something like saying that a large number of male students wake up with headaches, are still wearing their shoes so you wonder if the headaches are caused by wearing the shoes but we know that isn't true.  It is a correlation, not a causation because wearing shoes do not cause headaches. 

The next situation is where the numbers used imply the data comes from a random sample but in reality it came from a non-random sample. An example of this, is when the data comes from an online voting site. The data from one of these are not random because the people who participate tend to read the website that is hosting the survey and that means only certain members of the whole population are involved, not a good cross section of the population. 

Then there is adjusting the visual so it "looks" as if the data is supporting the hypothesis.  This happens when they change the values of the y-axis so the differences look bigger or smaller than they actually are.  The last way is to choose the lowest and highest values to be used in the analysis of the data.  We often see this type of adjusting with the top 10 lists, or the top 5 ways, etc.  It is not important for many things but if you do it for say the top 10 colleges, the information used to determine this list may create issues for readers if not done correctly.

Contrary to the character's line, numbers can and do lie based on it's presentation to people. Let me know what you think, I'd love to hear.  Have a great day.