Elizabeth Heyman points us to this display by Adam Pearce and Dorothy Gambrell who write, “We scanned data from the U.S. Census Bureau’s 2014 American Community Survey—which covers 3.5 million households—to find out how people are pairing up.” They continue:
For any selected occupation, the chart highlights the five most common occupation/relationship matchups. (For example, male firefighters most often marry female nurses, while female nurses most often marry managers.) Same-sex occupation/relationship matchups weren’t common enough to reach the top five in any occupation. So the chart also highlights the top male-male and female-female job matchups for each occupation.
I find the graph nearly impossible to read. But the data are fascinating. It could be a good class project to display these data in different ways.
As is often the case, the biggest contribution is not in the graphic itself but in the idea of going through the data to display this information in the first place. What a great idea! I look forward to seeing many different displays of different aspects of these data. And, thanks, U.S. government, for providing this information for us.
Fascinating. It would be nice to see some sort of adjustment for relative frequency. Almost every occupation on the male side has lots of marriages to teachers, secretaries, and nurses, presumably because there are lots of teachers, secretaries, and nurses. (The male equivalent appears to be truck drivers and retail supervisors.)
It would also be nice to see some sort of significance indication for the low count occupations. Do ballet dancers really tend to marry welders?
David:
Yes. The analysis again is useful not so much for what it says but rather as a motivator for researchers to play around with these data and see what’s happening.
Didn’t you ever see Flashdance… Ballet dancers are all welders
If I’m reading the visualization correctly, male gay writers marry editors, while female gay writers marry pharmacy aides? (It’s a thin line, though.)
If only they had done this in a more obnoxious and Facebook-friendly way, like an unreadable heat map or a word cloud for each occupation/gender combination. Then you stick that into a Prezi and you publish in the biggest open-access journal in the world: public opinion.
What’s the best (or just a better) way of visualising such data?
For starters, is the data set available somewhere?
I guess one could go back to the raw census dataset but that’d be quite the chore to prune & clean up & extract only the fields of interest.
How do they get the spousal occupation? The data dictionary for the American Community Survey has a respondent-occupation field OCCP but I cannot find spousal-occupation in there.
Anyone know?
http://www2.census.gov/programs-surveys/acs/tech_docs/pums/data_dict/PUMSDataDict13.txt
I was wondering the same thing. I have never seen that field in the ACS data and don’t know how to find the data they used in this study. Anybody know?
You need to use the relationship variables (e.g., HHT, RELP) to figure out the relationship between household members. Occupation is provided for each household member over the age of 15. So if person 2 is person 1’s spouse, then you can look at OCCP for person 1 and OCCP for person 2 and know the occupations of each spouse.
I’m did a data visualisation course last term and used this as an example of a bad visualisation in a report :)
Ernst Gellner: Social structure is who you can marry, culture is what you wear to the wedding.