Graphs and Stats

Seeing as I work for a company that does statistics, you’d think I would be really good at it. Unfortunately, I’m not. It is an entire world of math and formula that I have limited experience with have yet to have requirements that force me to acquire that experience. What I do see pretty consistently are graphs. I can say with the utmost confidence, that graphs are not what statistics are really about.

A graph is all about communication. It is a visual representation of some data that is meant to convey some relationship or lack of a relationship. Graphs can be changed and manipulated to express differences as huge jumps or radical contrasts. They can be used to reveal real trends and find outliers. A graph is just a tool used to communicate, which means it can be accurate or biased.

Statistics on the other hand are math, plain and simple. They are formulas and theories that help take answers and calculate relationships with specific understood precisions using specific constraints. From the outside it looks like statisticians looking at graphs and giving their thoughts, but really it is calculating values and adjusting inconsistencies in order to put a number on some trend. This was a tough thing to understand when I took stats in college. The class was really all math and we were told very specifically how to phrase results. The reason being is that stats is not about predicting or finding correlation, it is about assigning values to results much like equations are used to map things like sound waves and graphics.

I mention this because recently I’ve seen some graphs on social networks with commentary on the results as though something has been proven correct. One thing my experience at my job has taught me is that statistical data doesn’t prove anything. When I see graphs I appreciate the communicative aspects,  but rarely do I trust the conclusions. I’d rather see the stats.