This is my analysis of the claim that some fonts are more credible than others (Hacker News discussion here). That article claims that Baskerville has a small but statistically significant advantage in credibility over Comic Sans. It says, "The p-value for Baskerville is 0.0068." I'm not sure what that means, but this is my effort to analyze and understand the data myself. I'm not a statistician, but I'm trying to learn statistics, and doing my own analysis seems like a good "homework assignment."
The data set is provided in the file
raw. I use a Ruby script,
build-ad, to put the data into a useful format for R, stored in
ad. (You can run
build-ad by typing
make in the repo's top directory.) Some of my R commands are recorded in
explore, so you can see what I'm up to.
I'm just an amateur, so comments are welcome! I'd love to learn something.