rollin:
You have to collapse the serials and multipart works to singles so that you treat each as a single work or it penalizes the author who likes to break up his/her story into multiple small pieces.
Except that there are readers who will comment on, if not every part of a serial, on more than one. So those authors 'score' higher. Also, some stories may be favorably received, but since they are read by fewer readers (due to a poor title, or whatever), a high percentage of favorable responses may not translate into as high a number of total reads and comments; an author with a higher percentage of such stories, may appear to have a lower number. Bottom line, there is a problem with whatever metric you might try to devise.
Also, this doesn't really answer the question by dan2bend, which was about CHARACTERISTICS of stories, not AUTHORSHIP of stories. I'm a big stats person myself (my eyes didn't roll at all at dan2bend' mention of statistics), but I wonder what the point is when we have a great bunch of stories here. How about we just read them?