, , ,

Being an Econ major I read a lot of academic articles dealing with regressions.  Usually I gloss over the details in the methods section and do not worry too much about the structure of the regression and the other math involved.  However, now that I am in my methods course and learning about regressions I am taking the extra time to try to decipher these previously untapped sections, and I find it quite insightful.  It changes article to article, but in general I find I am taking much more out of my readings now.  Of course some setups remain a bit vexing due to their complexity, but overall I am definitely making progress.  I really wish this was taught a lot earlier in the major…

Anyway, something I realized while reading tonight is the importance of differentiating between statistical significance and significance as we use in everyday speech.  Normally we call something “significant” if it is important.  It has a meaning, it is relevant, and it is not inconsequential.  But in Economics (and any other discipline involving statistics) “statistical significance” has a very different meaning.  It does not measure the intensity of an estimated result but rather  the likelihood that the result comes from a relationship between the variables.

If statistical significance is low the estimated result is likely a cause of chance rather than an actual relationship between variables.  If it is high chance is likely a small factor.  Thus we say “with 95% confidence.”  This means that we can assume with 95% probability that our results are not products of chance.

This is elementary and every Economist should know this.  Actually, every academic should know this.  Somehow this has managed to slip under my radar…I wonder if it is my own failing or a trend my peers face as well?  If I am not a fluke we should consider emphasizing this more in our earlier classes, along with the other basic information to understand regression analyses.  I know that I at least would have appreciated understanding all the articles I have read a little more thoroughly…