Things some people wear round their necks. Hm, no, not those. This refers to the situation in which you might be interested in the rank order of some values but some have exactly the same value.
Details #
Say you have people whose ages are 30, 48, 19 and 57. Then the ranks of their ages are 2, 3, 1 and 4 as 30 is 2nd in that small set, 48 is third, 19 is first and 57 is last.
However if you add one more person and now have ages of 30, 48, 19, 57 and 48 then you have a tie in your ranking. You might say the ranks now are:
2, 3.5, 1, 5 and 3.5 if you use mean rank for ties
2, 3, 1, 5, 3 if you use minimum rank (think of ties in a competition) or
2, 4, 1, 5, 4 if you use the maximum rank.
For typical therapy/MH data mean ranks probably make most sense.
Ties only really matter to us because they introduce complications when computing statistics based on ranks such as the Spearman or Kendall correlation coefficients or other non-parametric statistics.
Try also #
Spearman correlation coefficient
Kendall correlation coefficient
Non-parametric statistics
Chapters #
Not specifically mentioned in the book.
Online resources #
None
Dates #
First created 27.xi.23.
