Update: Over a day later I realised that the render was taking the wrong RMD input. Sorry for time wasting.
I have been trying to run a series of parameterised Rmarkdown reports, but am finding that the report does not produce consistent outputs when using the "Knit" button in the Rstudio IDE vs when I generate the report using the rmarkdown::render() function (whether in a purrr::walk or as a one-off call to render(). In the latter case, tables contain strange bugs, e.g. a date column is correctly output, but the associated counts and percentages in the columns are duplicated down the rows.
I would provide a reprex but I can't share the report and underlying data (proprietary) and I can't seem to reproduce it in a reduced simulated data state. I have cleared caches and removed files, but am at a dead end on how else to problem solve this.
Can anyone advise what might be causing these differences between these two approaches or how I might get closer to working this out?
Thanks in advance.