Romney can still hack it where unemployment is low
As several media outlets have noted, both Iowa and New Hampshire have low unemployment rates compared to the rest of the country, possibly undermining the Republican candidates' perseverating claim that they are better equipped than President Obama to create jobs. Conventional wisdom holds that Romney and Paul should fare the best in places where unemployment is highest, since the economy is so central to their campaigns.
Here at the Signal, we crunched the numbers from Iowa to see if this held true last Tuesday. The color of each of Iowa's 99 counties here corresponds to the winner, while the opacity, or shading, represents unemployment. In other words, darker counties have higher unemployment rates, according to the Bureau of Labor Statistics' monthly local area unemployment figures.
If Romney and Paul are indeed hobbled by good jobs numbers, you'd expect to see darker green counties and blue counties and lighter red ones. Eyeballing the map, this does seem to be the case. But no map is perfect, of course, and this one doesn't represent the margin of victory in counties, which was often very close, or the volume of votes. (I tried to use intermediate colors, which looked like a very messy painter's palette.)
When you get down to the numbers, you see that Romney in fact has a 0.12 correlation to the unemployment rate, while Santorum has a -0.23 correlation, meaning he fares worse where jobs are scarcer. To put this in perspective, a 1.0 correlation would mean the unemployment rate would exactly predict the winner—something we obviously didn't expect given the complexity of the caucus system. In fact, a 23 percent correlation is fairly significant.
When we shared the figures with our Yahoo Labs colleague and stats-wiz Sharad Goel, however, he pointed out that the correlation is almost completely driven by the two northwest-most counties, Lyon and Sioux (labeled here), both of which had very low unemployment rates and went for Santorum. If you ignore those two counties, the correlations drop significantly.
We'll run the same figures on New Hampshire, though it only has only 10 counties. The real test of this factor will be in South Carolina. Iowa and New Hampshire both had a 6.1 unemployment rate for calendar year 2010, while South Carolina's was 11.2. (Note that these figures are calculated differently from the national figures, which are based on surveys impractical at a local level.)
