There are many issues when determining a mode.
1) for discrete values:
a) as Doc says, what about multiple modes, two or three values with 2 occurrances, or 2 values with 3 occurrances?
b) what if all the values are unique?
2) data grouping:
could you have 1,2,3 10, 15, 20 as the values?, then there is no discrete mode, but there is a clumping, so the "mode" would be the 1,2,3 group, which could be expressed as either "2" or "3" (midpoint or upper boundary).
Similarly, 1, 5, 10, 15, 18,19,20, would put the "mode" at the other end of the dataset.
And 1,2,3,18,19,20 would be bimodal, after a fashion.
Message was edited by: Chuck