Per Capita Alcohol Consumption versus Temperature by State

by robekulick

So I’ve been busy studying for a test this Saturday (pretty evil of UMD to have a test on a Saturday, but hey I guess such is the life of a PhD student), but as promised here is the regression analysis of per capital alcohol consumption versus gdp with another variable included controlling for the average temperature in the state as reported by the NOAA. Note that for some reason the NOAA omitted temperature data for DC, Alaska, and Hawaii, so the sample size is 48.

First here is a link to my previous post on this subject:

https://robekulick.wordpress.com/2011/08/20/more-economic-fun-alcohol-consumption-per-capita-versus-gdp-per-capita-by-us-state/

And the results:

Linear regression                                      Number of obs =      48
                                                       F(  2,    45) =    6.05
                                                       Prob > F      =  0.0047
                                                       R-squared     =  0.1521
                                                       Root MSE      =  .17899

------------------------------------------------------------------------------
             |               Robust
   lnalcohol |      Coef.   Std. Err.      t    P>|t|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
       lngdp |   .2819775   .1249402     2.26   0.029      .030335      .53362
     lnjtemp |  -.1002307   .0507621    -1.97   0.054    -.2024708    .0020094
       _cons |  -1.803124   1.382976    -1.30   0.199     -4.58858    .9823324
------------------------------------------------------------------------------

As I suspected would be the case, the statistic for January average temperature is negative and significant, while GDP per capita remains positive and significant. These results suggest that a 1% decrease in the average temperature in January in a state is associated with a -0.10% increase in per capita alcohol consumption. Again, it can’t be determined whether this effect is causal since there are a number of other variables I would need to control, but the association is pretty interesting in and of itself!