Central Tendency (Measures):
Any measure of central tendency provides a typical value of a set of 
values 
. Normally, it is a value around which values 
are grouped. The most widely used measures of central tendency are (arithmetic) mean , median , trimmed mean , mode . Measures of central tendency are defined for a population and for a sample .
For example, two samples - (8,9,10,11,12) and (18,19,20,21,22) have central locations differing by 10 units, and most measures of central location would give values 10 and 20 of the two samples, respectively.
Measures 
of central tendency normally meet the following requirements:
- If all values
coincide - i.e. all are equal to the same value 
- then the measure 
is equal to 
, or formally
- The value
is within the interval between the minimal and the maximal value of the set 
:
- Measure
has a property of shift invariance :
where
may be negative. - Measure
has a property of scale invariance :
Most measures of central tendency also have the following property: If the set of values 
is symmetrical with respect to a value 
(the center), then the value of the measure 
coincides with the center 
. "Symmetrical" here means that, for each value 
different from 
, there is another value 
deviating from the center 
by the same magnitude as 
, but in the opposite direction. (More generally, "Symmetrical" means that the right and left sides of a distribution look the same, in mirror image.)
 |
Note that some measures are often classified as measures of central tendency (and have "mean" in their names) but do not meet the requirement of shift invariance. Such measures are usually defined mathematically only for non-negative values 
and, practically, are applicable to quantities that are non-negative in principle - e.g. price, time or space interval, weight, etc. Strictly speaking, such descriptive statistics measure "effective magnitude" or "average magnitude" rather than central tendency. Some examples of such measures are: the power mean , the harmonic mean , the geometric mean , root mean square .
See also Mean Values (Comparison)