More fun with 1.usa.gov data
I took another look through the 1.usa.gov data, this time to see if there is a difference in the distribution of browsers between the different top-level domains (nasa.gov, fbi.gov, weather.gov, etc.).
First let’s take a look at the overall browser usage for the data that I’ve collected.
You can see that Internet Explorer is the most popular, followed by a near tie between Firefox, Chrome and iOS devices. This is very biased, considering that the actual distribution of browsers according to wikipedia looks like this:
Those mobile browsers include both the iOS and Android values from my graph. According to bitly, 22% of their links come from twitter or facebook, and since those sites are so often viewed on mobile devices, it makes sense that mobile browsers are over-represented. It also appears that Internet Explorer users are very under represented.
Since 47% of all of the clicks in my dataset point to nasa.gov, the distribution of browsers those links is of interest as well. Here’s what visitors to nasa.gov are using:
At the 1.usa.gov hackathon, I showed that people outside of America are primarily interested in nasa, where Americans are interested in a broader range of government websites. Since the market share for IE is much lower in Europe than in America, that partially explains why chrome and firefox are more popular than IE on nasa.gov links and less so on other government websites.
Let’s see how that compares to the next 11 most popular websites (neglecting shrewsbury-ma.gov).
This isn’t really all that informative, other than to give a sense of just how popular nasa is compared to the other sites. Here’s that same plot, normalized to show the percentage of clicks for each site rather than the total count:
It’s interesting to see that the usda, fda, and noaa websites have very similar distributions. It sort of makes sense that the same people will visit usda and fda websites, and I noticed at the hackathon that noaa is most popular in the Tornado Alley area. My next step is to mash the browser information with geographic data and see what happens.