This folder should contain the cleaned data sets. There should be one .csv
file (ideally) per experiment, and the files should adhere to the standard
naming scheme. If there is an experiment called "urntask", for instance, the
corresponding data set should be called data_urntask.csv.
The data files should be generated automatically, using the scripts in the preprocessing folder. They should not be edited by hand, nor should any of the analysis files alter these data files.
To the extent that this is possible, the format of a data file should adhere to the idea of "tidy data". Each row should describe a single observation (i.e., each response from the same person should appear in its own row), and each column should correspond to a single variable.
Variable names should use "snake_case", should not include white space, etc.
Generally speaking, the variable names should be the kind of thing produced by
janitor::clean_names() and should be easy to understand.
Values stored should be informative (e.g., store gender as text "male", "female",
etc, rather than numeric dummy codes 1, 2, 3). If something about the data is
not obvious from inspection, there should be explanatory "notes"
file, called notes_urntask.md. Unlike the CSV file, the notes file can
(and probably should) be created manually.