I thought shoe sizes and rack dress sizes were interval data, not
ordinal - the difference between a size 10 and 11 was the same as a 5
and 6 (some constant number of centimeters).  You're quite right that
the 0 point is frequently arbitrary, however this still makes means and
variance and such valid measures.

To defend my assertion of ordinality, consider a user who rates apps a 1
star if they crash instantly, and then if it works somewhere between 2
and 5 stars depending on quality, with 4 stars being "good" and 5 stars
being "favorite".  If there were a good chunk of users rating in this
fashion then it's very silly to claim that some percentage of users
promoting the app from a 1 to a 2 is the same as some percentage
promoting it from a 4 to a 5.

Put briefly, one star could be the equivalent of a size 0 dress on a
size 14 woman, while 2 through 5 stars were sizes 10-14.

-- 
You received this bug notification because you are a member of Ubuntu
Bugs, which is subscribed to Ubuntu.
https://bugs.launchpad.net/bugs/894468

Title:
  Statistics algorithm for sorting ratings looks fishy

To manage notifications about this bug go to:
https://bugs.launchpad.net/ubuntu/+source/software-center/+bug/894468/+subscriptions

-- 
ubuntu-bugs mailing list
[email protected]
https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs

Reply via email to