I’ve done some experimenting with “rapid response demography,” using Google tools to track time trends for marriage and divorce, recession and divorce and unemployment. Beyond time, we also could look at spatial variation, less for trends than for clues to cognitive, cultural, or subtle demographic patterns we weren’t expecting. Since my thinking is exploratory, and this is a blog, I’ll share what I found without attempting to draw conclusions. This uses the Google Correlate tool, a free tool (login required) described in some detail in this whitepaper.
With the new American Community Survey marital event data, for the first time we have recent divorce incidence data for every state (the weak vital records apparatus for divorce in the US is slower, inconsistent, and doesn’t include all states). I combined 2008 and 2009, and calculated the rate of divorce per 1,000 married women.
So, what were people searching Google for in the states with the highest divorce rates? Handguns.
This isn’t just a fluke with Colt 45. Out of the 90 terms most correlated – across states – with the divorce rate, 28 are about guns, and not the hunting kind but things like “kel-tech sub 2000” and “glock custom.” I don’t see a clear pattern in the rest of the list, though “splenda cake” and “low carb thanksgiving” are intriguing.
Way down the list, though, at #93, with a correlation of only .60, is “tea party.” After #90, Google doesn’t give you a state map, so I made this scatterplot:
The states where people search for “tea party” are also those where the divorce rate is higher. Which made me think of politics, and the simple correlation between divorce and the red-state-blue-state thing. Some writers have reported that red (politically conservative) states have high teen marriage/pregnancy and divorce rates, which is partly because of how much Evangelical Christians get pregnant and married at young ages.* But I haven’t seen a direct comparison with recent data. So here is the divorce rate with the Obama vote in 2008:
The states with higher divorce rates gave smaller shares of the vote to Obama in 2008. But this correlation isn’t so strong, just -.48 on a scale of 0 to 1. What if you put the Obama vote into the blender and ask what Google searches follow the Obama pattern? The answer is overwhelming: Top Chef:
Again, not a fluke. No less than 13 of the top 90 correlated search terms are about Top Chef (“bravo top chef,” “top chef episode,” etc.). Someone who has seen this show might have to give me some hypotheses for why Top Chef states went for Obama.**
Aside: Am I just cherry-picking results according to some preconceived agenda? No and yes. The green graphs are by Google, and they represent the #1 search correlations with the real-world state indicators I uploaded (divorce and Obama votes). The blue scatter plots are things that struck me as interesting, that I chose to pursue. That’s how exploration works: look at the pattern, dig around, repeat.
* Jennifer Glass and Philip Levchak presented a paper at the Population Association of American conference this year which analyzes divorce rates in all U.S. counties as a function of a variety of factors, including religious affiliations. The paper, which does not seem to have been published yet, concludes in part:
The results here show that communities with large concentrations of religious conservatives actually produce higher divorce rates than others. … The major pathway linking religious conservatism and divorce seems to be the tendency of conservative Protestantism to encourage the early cessation of education in favor of marriage and childbearing. [Although, they add, much of the conservative Protestant effect remains unexplained in their models.]
** Two nights ago, a Tweet from @MrBrwnIYZ: “Well ain’t this about a bitch… President Obama decides to do an address to the nation when Top Chef comes on!”