I posted recently about how the state-by-state unemployment rate has changed during my lifetime. The result was a small multiples grid that put the states in context with one another.
Today I’ve created a new version aimed at identifying more precisely how each state has differed from the national unemployment rate during the last four decades. The lines show the percentage point difference — above (worst) or below (better) — from the national rate.
This view allows us easily to identify the most anomalous states in both directions (West Virginia, for example, had quite an unemployment spike during the 1980s; South Dakota, on the other hand, has never been worse than the national rate).
There’s plenty more to explore in this quick remix:
There’s good news this week in the monthly jobs report, the latest sign that the economy, however grudgingly, has healed from the financial crisis nine years ago:
The unemployment rate fell to 4.6 percent, the Labor Department said, from 4.9 percent. The last time it was this low was August 2007. That was the month, you may recall, when global money markets first froze up because of losses on United States mortgage-related bonds: early tremors of what would become a recession four months later and a global financial crisis nine months after that.
These things, of course, are cyclical. Here’s how the unemployment rate has changed, by state, during my lifetime:
As FiveThirtyEight notes, turnout in the 2016 presidential election isn’t dramatically lower than it was four years ago, according to the latest estimates. And with many mail-in and provision ballots still being counted, the 2016 turnout rate could still change:
Approximately 58.1 percent of eligible voters cast ballots in last week’s presidential election, according to the latest estimates from Michael McDonald, associate professor at the University of Florida, who gathers data at the U.S. Elections Project. That’s down only slightly from 2012, when turnout was 58.6 percent, and well above 2000’s rate of 54.2 percent. Turnout may end up being higher than in any presidential election year between 1972 and 2000….
We won’t have final turnout numbers for weeks or months because some states are still counting ballots; millions remain uncounted. That means estimates based solely on votes counted so far will understate turnout — though already more presidential votes have been counted this year than in 2012 (contrary to reports that fewer voters turned out this year). In the meantime, most news organizations rely on estimates from McDonald.
Here’s a quick look at historic turnout in both midterm and general elections, according to estimates compiled by McDonald:
Back during the Republican primaries, The Upshot published an interesting short post called the Geography of Trumpism. The reporters back then analyzed hundreds of demographic variables, by county, in an effort to determine which ones might be predictive of electoral support for the eventual GOP nominee.
Think: What’s the rate of mobile home ownership? Or what percentage of people in a particular place have college degrees? They found a key variable to explore:
When the Census Bureau asks Americans about their ancestors, some respondents don’t give a standard answer like “English” or “German.” Instead, they simply answer “American.”
The places with high concentrations of these self-described Americans turn out to be the places Donald Trump’s presidential campaign has performed the strongest.
I’ve plotted the percentage of “American” ancestry, by county, on a national map. Keep in mind the data come from a five-year survey by the U.S. Census Bureau, so the accuracy in large counties is relatively safe.
But in smaller counties — say, those with fewer than 10,000 residents — the margins of error can be quite high. The results are even more problematic in the tiniest of counties. Still, this is the best public data we have, and it does produce some interesting geographic trends:
Earlier I used small multiples to show how each Major League Baseball team’s 2016 season progressed relative to the .500 line. Here are those same line charts, but this time I’ve grouped them by division:
Last week I published a new heatmap exploring the popularity of American birthdays. The chart, which uses darker shades to represent higher average birth counts on specific days, can give the impression that some birthdays are much more common than others.
In reality, outside of some special occasions, namely major holidays, there isn’t a huge amount of diversity in the data set, which has two decades of births aggregated by day. Most birthdays, including my own, are fairly average — especially in the first six months of the year. For example:
It’s baby season in America, with September the busiest month for births on average in the last two decades. So it seemed like the right time to remix this blog’s most-popular post: How Common is Your Birthday?
That old heatmap, which highlighted specific dates for popularity, has been viewed more than 500,000 times here and published across the web. But it was flawed, namely that it used ordinal data (birthday ranks by date) rather than continuous data (actual births counts by date). This graphic finally addresses that problem:
Note: I followed my wife, a foreign correspondent for NPR News, to Seoul last year. This is one of a series of posts exploring our adopted country’s demographics, politics and other nerdy data stuff. Let me know if you have ideas for future posts.
I’ve been away from Seoul for much of the summer, but now that I’m back it’s impossible not to hear all the complaining — among expats and locals alike — about the heat.
They have a point, at least in terms of their expectations. This summer has indeed been hotter than usual, especially this month, when the daily low temperature on one recent day actually exceeded the average high. (I updated the chart on Aug. 24).