What’s in an Average?

Someone once said about data: “there are lies, damned lies, and statistics.”  This truism reflects the fact that information gleaned from data can often be maligned.  Here’s a good example.

The Mercatus Center at GMU produced the Freedom Index of the fifty states.  The fine folks at George Mason took a plethora of statistics regarding personal and economic freedom from each of the states and reduced it to a single Freedom Index score.  Part of that data set contained information regarding right to work (RTW) laws.  I wanted to see if there was correlation between RTW laws and statewide employment.  My intuition told me that, in general, unemployment should be lower in states with right to work laws.  Forced unionism can be an impediment to employment, or so I thought.

I went to the Bureau of Labor Statistics and obtained unemployment figures for each state.  I used the most recent data – January 2013.  This is a snapshot of a point in time and a better indicator might have been a two or three year average.  But to keep it simple, I used the January data.

First, I ran a simple average for RTW states and the unemployment rate. If my intuition were correct, I should find lower UE in RTW states.  Here’s what I found:

                Unemployment rate for RTW states – 6.78%

                Unemployment rate for non RTW states – 7.3%

Unemployment is over 7% lower in RTW states.  Case closed, right?  If I worked for the MSM, I could conclusively report that states with RTW laws have lower unemployment figures than non RTW states.  Therefore, RTW laws increase employment.  Not so fast.  While the average UE rate in RTW states is indeed lower, is this relationship strong?  Let’s see.

Using unemployment as the dependent variable and RTW law as the independent variable, I used regression analysis to see how RTW might influence employment.  My R-square value was a mere .027 and the P-value was not significant at 0.253.  What does all this mean?  While the average unemployment rate is lower in RTW states, only 2.7% of the variation can be explained by RTW laws.  Other factors are at play here and RTW alone isn’t sufficient to explain why RTW states enjoy lower unemployment.

Moral of the story: beware of misleading statistics. We are constantly bombarded with averages, percentiles, and other data regarding a whole host of economic data.  Many times, these statistics are meant to influence our thinking towards public policy.  So, before you hop on board your favorite statistic, look more deeply into the numbers.  You might be surprised at what you find.