Sorry but I can’t follow you here… Of course the average does not need to match one of the scores (why would it anyway?), it rather conveys information on the overall distribution of opinions, and that’s what I’d like to capture. An example: The median of [1, 1, 1] is the same as the median of [1, 1, 5], namely 1, so it totally swallows the information that there was one person who thought the book is super-hard. Using the average calculation you get 1 vs. 2.33 which reflects this distribution of votes much better in my opinion.
Why do you think the average doesn’t make sense?
Not necessarily; the median of [1, 2] would still be 1.5, no?




And since it’s a new addition, the example pages for this are much fresher in my mind than the other books.
We had 27 and 24 participants in the previous polls. It’s a bit sad to see that more than a quarter of the voters dropped out. But maybe that’s just life 



