During every Presidential season, all the various news agencies seem to like to go to what they call a “bellweather county.” This is described as “the only county” which has consistently voted for the winning candidate in every Presidential election since some date in the past (with occassional exceptions). The news outlets all treat this as some kind of remarkable feat (and I’ve noticed that the counties tend to change from election to election). Am I wrong in thinking that this is total BS on the news agencies part? That statistically speaking, given that most Presidential elections are decided upon a majority of the population voting for one particular candidate, finding counties with a long history (say 100 or so years) of “picking” the winner really isn’t all that difficult, and that there’s a lot of them out there?
I realize that there’s no way a place like Alaska, Hawaii, or Arizona (to name but a few) could have a history going back so far, but in some place like New York state, I’d think it’d be fairly easy to find a county which voted for the winning side in all but one or two Presidential elections.
The word is actually “bellwether”: a bellwether is a sheep {often a castrated male, or wether} with a bell around its neck which the rest of the flock would follow.
There’s a nice NPR story on this subject available on their website. Apparently there are six counties who have voted for the winner in every election since 1960: Vigo, IN; Lincoln, MO; Van Buren, AR; Logan, AR; Eddy, NM; and Ferry, WA. Vigo County has voted for the winner in 26 of the past 28 presidential elections (since 1892.)
The United States Geological Survey (USGS) says that there are 3,141 counties and county equivalents in the United States. Assuming an average margin of victory is around 60% of the counties going to the victor, some have been closer and some have been landslides, of course, but 60% is pretty close, if the odds of choosing the winner were completely random, one would expect 0.6^12 counties to have selected the winner 12 presidential elections in a row.
That works out to be 0.002177*3141 counties=6.8 counties getting it right 12 in a row, just slightly more than the six that did. There’s no bellwether effect, it’s just an effect of large numbers, repeated trials and ignorance about statistics.
I hope it’s OK to resurrect this thread; I was curious about whether any of the above counties had retained bellwether status after the 2008 election. Believe it or not, only Vigo, IN went for Obama; the other five went for McCain.