Hi @valosekj, I’m so glad you asked! I’ve prepared my fair share of phenotypic data and I suggest you think about this from your end user’s perspective. While it may be convenient to you to put many surveys in one TSV with one data dictionary JSON each for the two “varieties”, I am a proponent of keeping them aggregated at the top level BIDS “root” folder’s phenotype/ subdirectory where they should be separated by their smaller unit, as you called them “the individual files”.
So I guess the short answer is “individual files” so an end user doesn’t have to sift through a large TSV and can instead go right to the phenotypic data they care about more directly by a more logical separation. It’s late here, but does this answer make sense?