I mean... the nature of R stats was never that fitting a model was hard work, but making sure it's the right tool applied correctly and making meaning out of it is what makes it powerful. But it does make me fear what unsupervised mistakes might lurk in the mechanics like data wrangling 🙈

We worked with an Excel "form" at first, but its encoding created more problems. We recently switched to a platform that allowed us to do some pre-validation + if-then field requirements. tally.so/r/w7jGe2

SQL was also a pain to deal with in terms of encoding. If the database will be small enough that you don't benefit that much from the querying power of SQL, I would just go for well-structured CSVs with a lot of linking keys/identifiers.

there are quite a few packages that make interacting with SQL easier through R like MariaDB and especially dbplyr for tidyverse like querying. I work with BioTIME and we struggled with SQL and R handle string recognition differently (e.g. giving different totals for distinct species).

thanks for replying so quickly!!

might be a silly question, but does Proc B compensate data editors or is this a voluntary term?

latex default is pretty painful

what do you have against old typesetting :o I'm personally not a fan of the sans serif Wileys or soulless Elsevier layout