How do people tend to analyze log files? Are there any packages out there to do so, or do people tend to resort to custom scripts each time?
Hey @rkempner, lovely question!
Personally, I haven’t run that many studies via PsychoJS/Pavlovia yet, but with systems I developed in the past, I have also developed associated packages/frameworks. It often didn’t really have a lot of added value though; if you build something flexible enough to deal with all the different data structure a system like PsychoJS can produce, you risk just building R/dplyr again :). See Inner Platform Effect.
On a different level I have built something that might be useful: the splithalfr package (preprint) for calculating split-half reliabilities. To prevent the Inner Platform problem, I modeled my functions after base R functions (apply).
I see, I was naive to the fact that R can actually handle log files.
Thanks! Thanks for the link to Inner Platform and your package.
You’re welcome! It can as far as I know, but the data wrangling can vary a lot per task. It’s quite a challenge to build something more elegant and powerful than what R and standard packages already offer. If you’d have a specific task that’s used often, then it might pay off to standardize data structures and build something for that.