This nice paper by @JZBerman, @deborahasmall and colleagues is relevant (1 to 7 scales where 7 is max agreement)