I think one the best solution around here is to use non-relative pass, and get helps for that from this package
You could write
war <- read_csv(here::here('data/warriner_2013_emotional_valence.csv'))
and that will generate a full path to your file base on the root of your project. This way the R chunk will always managed to read your csv file no matter where the Rmd file is.
If knitting goes well, it means the option is working. If you run the chunk with csv reading on its own, you need to be sure the setup chunk is run before. In RStudio IDE, the green arrow to run chunk should insure that. Are you using RStudio ?