Using replication files I downloaded here, I run some static code analysis using lintr.
For R files in the replication archives of AJPS articles posted to Harvard Datavaerse, the statistics are somewhat grim.
> median(lint_results$style, na.rm = T)
[1] 156
> mean(lint_results$warning, na.rm = T)
[1] 5.866562
> mean(lint_results$error, na.rm = T)
[1] 0.005756149