From Nathan Yau. I love this stuff. It’s just wonderful, a great set of visualizations on a great topic.
Offhand, the only suggestions I have are to scale the graphs or indicate in some way the trends in the total popularity of each name (as it is, I wonder if some of the variation is arising from rarity), also to me the girl color looks a bit orangish and I’d go for something more purely pink.
P.S. These graphs are pretty good too.
“the girl color looks a bit orangish and I’d go for something more purely pink.”
That reminds me that graphs of demographic data these days tend to go out of their way to use arbitrary colors that are hard for the user to remember. It’s not uncommon to see blacks represented by, say, green, Hispanics by red, American Indians by blue, Asians by purple, and whites by brown. Other graphs will then use another intentionally confusing set of colors.
Is there research on the best shades to use for pink/blue contrast? The default pink/blue colors used in the lattice R package are totally indistinguishable to one of the main groups of the color blind, as I have learned the hard way. Perhaps Nathan intentionally chose more color-blind-friendly shades?
“I wonder if some of the variation is arising from rarity”
It seems like this is yet another example of where some kind of shrinkage would help. Look at the end of the graph for ‘Gale’ for example. Mainly noise, which could be helpfully smoothed away.
Here’s a post on almost the identical topic published a month a half ago. I wonder if Nathan is getting some inspiration without citing his sources.
http://betweenthenumbers.net/2013/08/what-is-the-most-unisex-name-in-us-history/