Below is the single most extraordinary chart explaining the results of the recent Presidential election:
It’s widely assumed in the press that victory in the Electoral College [2] is determined by the Gender Gap [3] or by the Rising Tide of Hispanic Voters [4] or whatever. But in fact the relationship between these demographic factors and whether a state votes Republican or Democratic in the four Presidential elections of this century has been relatively weak.
Despite the increasing importance of nonwhite voters [5], what still [6] determines Presidential elections is a fundamental divide among whites over the very basics of life. Thus, an extremely obscure statistic measuring marriage among younger white women that I debuted [7] here on VDARE.com in December 2004 correlates sensationally with Electoral Votes.
This metric: average years married among white women ages 18 to 44 on the 2000 Census (what I’ll call “Years Married” for short).
“Years Married” had its best won-loss record yet in 2012. Mitt Romney carried 23 of the 24 highest-ranked states. Barack Obama won 25 of the 26 lowest-ranked states.
In my chart above, the length of each state’s bar indicates the average number of years that a white woman 18-44 can expect to be married. Romney’s states are colored in the now traditional Republican red and Obama’s in Democratic blue, with Romney’s share of the two-party vote next to the name of the state.
At the top of the chart is Utah, where white women average 17.0 Years Married and Romney won 75 percent. At the bottom are Massachusetts and California.
In Massachusetts, white women average only 12.2 Years Married and Romney was beaten roughly 5 to 3.
(I left off the District of Columbia [8], a nonstate that gets three Electoral Votes. [9] White women only average 7.4 Years Married there, and Romney won merely 7 percent in the capital.)
The sole anomalies were Obama capturing Iowa (which is 21st in Years Married) and Romney taking Arizona (41st).
Republicans need to ask themselves seriously why they didn’t win Iowa [10]. Don’t ask: “What’s the Matter with Iowa?” Instead, ask: “What’s the matter with the GOP that they can’t win a respectable state like Iowa [11]?
As for Arizona, I suspect that it’s culturally an exurb of Hollywood [12], but politically it’s an exurb of Orange County [13]. This may help explain the virulence of the New York Times’ long-running war on Arizona [14]: the Grand Canyon state is supposed to turn into California Jr., not into something new.
More on this Years Married statistic: please note that it is not a measurement of white people getting married in that state. Otherwise, Nevada [15], with its 24-hour wedding chapels [16], would be near the top of the list.
Years Married is a measurement of white people being married. Thus, states with high rates of both marriage and divorce, such as Oklahoma [17] (unofficial state song: George Strait’s “All My Exes Live in Texas [18]”), don’t perform quite as well as stable Utah.
The best predictor of Republican performance isn’t the rate of getting married—because if you have a state where a lot of people get married and then they turn around and get divorced [19], that doesn’t do the Republicans as much good. Divorced white people vote Republican less than 45 percent of the time, while over 63 percent of married white people go GOP. In short, Republicans do well among people who get married and stay married.
To demonstrate how stunningly sorted into red and blue this Years Married graph is, let’s compare it to a more celebrated demographic statistic: Percent Nonwhite.
Lately, everybody has been talking about how the growing nonwhite share of the population hurts the GOP (Peter Brimelow [20] has been talking about this [21] for 16 years [22]).
And, that’s true. It does.
Yet, when you graph it out state by state, the red-blue divide isn’t as clear. At the top of the chart are the whitest states, Maine and Vermont [23], which Romney lost in landslides. Then West Virginia, New Hampshire, North Dakota, and Iowa. Not much of a pattern.
If you look at this chart very carefully, you can see a little more Republican red toward the top and a little more Democratic blue at the bottom…but nothing like the Years Married graph.
This is not at all to say that Percent Nonwhite is unimportant to election results—just that the real world is complicated. Percent Nonwhite correlates with Romney’s share of the vote in the fifty states at -0.32: a low to moderate negative correlation. (It’s stronger if Washington D.C [24]. is included.) This is the kind of hodge-podge you normally see when you graph a single factor that impacts voting.
As I pointed out back in 2000 [25], having a lot of blacks in a state tends to drive whites to the Republicans [26], while having a lot of Asians seems to make whites more liberal, with Hispanics in-between.
You might think, reasonably enough, that this 2012 Years Married result must be a one-time fluke. Maybe it was the result of the personal characteristics of Mitt Romney, a philoprogenitive Mormon [27] who had 43 years married and 23 descendants? [28]
But in 2008, John McCain [29] carried 19 of the top 20 states on this same metric, while Obama captured the 25 of the bottom 26.
Here’s the 2008 chart:
As I said, I discovered Years Married right after the 2004 election. Unsurprisingly, it worked really well then, too [30], with George W. Bush winning the top 25 states and John F. Kerry [31] 15 of the bottom 18:
And, finally, there’s 2000, when Bush took the top 25 states and Gore 14 of the bottom 17.
You may be wondering: Why white people? Why not measure Years Married for everybody? The short answer is: that’s what works best.
There is a higher correlation with the Republican candidates’ performance and Years Married among whites than with Years Married among the entire population. Conversely, Years Married among whites correlates better with the GOP’s share of the overall vote than with its share of the white vote.
You could accuse me of “data dredging [32]”—trawling through many possible correlations looking for whatever turns out highest by random luck. But when I dreamed up my Years Married statistic in late 2004, it worked amazing well for two elections. It didn’t project all that well farther back into the past due to Ross Perot’s [33] third party runs in 1992 and 1996. But it has since worked well for two subsequent elections. What more can we ask of a statistic?
It took me a long time to find the single measurement that best correlated with voting by state. Way back in July 2000, I noted [34] here at VDARE.com that the most liberal state, Vermont [35], had the lowest total fertility while the most conservative state, Utah [36], had the highest [28]. After the 2000 election, I explained for UPI [37] that Bush had carried the 19 states with the highest white birthrates.
Immediately following the 2004 election, I pointed out in my “Baby Gap [38]” article The American Conservative that the “total fertility rate [39]” among whites was an uncanny predictor of overall voting by state.
Over the next couple of weeks in late 2004, I worked out how to create an age-adjusted measurement of being married. I reported in VDARE.com in “The Marriage Gap” [7] on December 12, 2004 that the rate of being married among white women 18-44 in the 2000 Census had the highest single correlation with voting GOP.
I then fortified my theory by including the impact of geography on home prices—“The Dirt Gap [40]”—which, in turn, determines the “Mortgage Gap [41].” In places where family formation is more affordable, the “family values” party does better.
Recently, political scientist George Hawley [42] of the University of Houston has confirmed my state-based theory at the county level in a study published in the academic journal Party Politics: Home affordability, female marriage rates and vote choice in the 2000 US presidential election: Evidence from US counties [43]. [February 24, 2011]
After a lengthy review of academic articles on voting, Hawley writes:
The possible relationship between home affordability and aggregate voting trends has largely been ignored up until now by the political science literature, though the topic has been considered by the political journalist Steven Sailer (2008 [44]). Sailer hypothesized that “affordable family formation”—which he argued was closely related to housing costs – was a key difference between majority-Republican states and majority-Democrat states. Sailer went on to conclude that the relative affordability of housing accounted for the differing typical political behavior within various large cities. Sailer suggested that the relative costliness of owning a home in America’s large coastal cities, such as Los Angeles, led to later family formation, which partially explained the greater support for Democratic politicians in those cities and regions. In contrast, inland American cities like Dallas are able to expand outward all-but indefinitely, which keeps housing costs low and subsequently [makes] such cities more attractive to young families.
Hawley went on to find a statistically significant effect at the county level in the 2000 election. In all likelihood, other scholars could find similar county results in the three subsequent elections. (In other words, if you are an academic social scientist searching for an important result to publish, check out 2004, 2008, and 2012.)
It’s worth looking at my scatter plots of the correlations. First, here’s Romney’s share of the two-party vote on the vertical axis vs. the 2002 total fertility rate for white women on the horizontal axis. The isolate in the lower left corner is Washington D.C., while Utah is in the upper right corner.
The federal government doesn’t go around calculating for every state, total fertility for white women very often, so I’m using 2002 numbers. They’re ten years out of date, but this is still a pretty good correlation: including D.C., the correlation coefficient is 0.83.
Psychometrician Linda Gottfredson [45] likes to say that in the social sciences [46], a correlation of 0.2 can be thought of as “low,” 0.4 as “moderate,” and 0.6 as “high.” So, anything above 0.8 must be very high.
The scatter plot for Years Married is even tighter.
The correlation is 0.88.
If we exclude Washington D.C. as an outlier [47], the correlation is still 0.84.
The correlations between the 2000 Years Married and the GOP's share of the vote has been very high since 2000. Including Washington D.C. pumps up the correlations to stupendous levels, but they’re still jaw dropping without D.C.:
Let’s explore the interplay of marriage and fertility for each state, calculating a ratio versus the national average of 100. It’s not easy to put each state on a graph, so I’ve come up with a Bar Chart Without the Bars that just displays the numbers in the appropriate locations.
Yes, that sounds odd, but check it out and I think you’ll see that it’s usable. The states are sorted in descending order of Romney’s share. Romney’s best state was Utah, where the white Years Married rate (dark type) is 121 percent of the national mean and the white Total Fertility Rate (red type) is 134 percent.
It’s easy to spot that the only states with exceptionally high white fertility are Mormon Utah [48] and empty Alaska. [49] Strong Republican states tend to have higher white marriage rates than fertility rates. This is especially apparent in deep southern states such as Alabama. Strong Democratic states tend to have low white fertility.
California, the leading prize [50] in the Electoral College [51], is down toward the bottom at 89% of the national white average for marriage and 90% for fertility. It is therefore not that surprising that what used to be the keystone to Republican success in the Electoral College, voting Republican 9 out of 10 times between 1952 [52] and 1988, has now gone solidly liberal, with Obama winning the white vote in California in 2012. )In the Narrative, California is always about Proposition 187 [53], but it had already switched to Democrat in the 1992 Democratic election, two years before).
And Republicans can’t exactly expect to carry any state where they lose the white vote.
California used to be the paradise for the common man [54]. Housing was no more expensive than in the rest of the country and the public schools were good. Inevitably, there was a huge influx from the other states, driving up real estate prices. But, quite evitably, there was gigantic illegal immigration into California, which devastated the public schools.
That all makes people wonder, well, if I can’t afford a house should we really bother to get married, and even if we can afford a house can we afford one in a good school district like Los Virgenes [55]? Or are our kids going to be stuck in classrooms overwhelmed by the children of illegal immigrants? [56]
And if we can’t really afford private school [57], or a house in an expensive school district, [58] then what’s the point of having kids?
And if we’re not going to have kids, what’s the point of getting married at all?
And if we’re not married, then don’t those Republican politicians get on your nerves with all their family values talk?
In summary, what all this suggests is, that rather than try to manipulate voting the way Bush and Rove [59] did through boosting subprime loans to Hispanics, [60] what Republicans need to study is: what makes white people get married—stay married—and want to have children.
I suspect that what the government can influence comes down to affordability: the cost of real estate; and satisfactoriness of public schools.
What the Years Married measure implies is that the people who vote Republican tend to be happy white people.
Therefore, it’s in the self-interest of Republican politicians to try to make white people happy.
Hint: Mass, non-traditional, immigration is not the answer [61].
Steve Sailer (email [62] him) is movie critic [63] for The American Conservative [64] and writes regularly for Takimag [65]. His websitewww.iSteve.blogspot.com [66] features his daily blog. His book,AMERICA’S HALF-BLOOD PRINCE: BARACK OBAMA’S "STORY OF RACE AND INHERITANCE", is available here [67] and here [68] (Kindle)