The Liking Initiative: a database of subjective evaluation ratings for decision-making research

Subjective liking ratings from published decision-making studies, collected into one database with a common normalized scale so ratings from different response formats can be compared. Each release ships the whole corpus as tab-separated files: a catalogue of every study and dataset, one file per dataset, and one file holding every rating. codebook.md documents the columns and the two things most likely to be got wrong — that cross-study comparisons must use normalized_rating, and that subject identifiers are unique only within a dataset. The likingInitiative packages for R and Python read these files directly. They are ordinary TSVs and can be used without either package. Every correction applied to the data is recorded in catalog.json as a numbered migration, so any value here can be traced back to the source file it came from.