Characteristics  of Good Average

a.It should  be Rigidly  Defined  An average should  be rigidly defined so that  there is  no confusion  in regard  to its  meaning  and connotation. If  it is not well defined it may be influenced by  the  prejudice  and personal bias of the investigator making its value  unstable and  render it unrepresentative of  the distribution.

b.Its Definition Should  be in the Form of a Mathematical Formula With  mathematical  formulation  different persons  may not interpret it differently and anybody computing the  average  from  a set of data will arrive at the  same figure i, e, it should  not depend  upon the  whim  or  caprice  of the person  computing it.

c.It Should  be Easy to Calculate and Simple  to Follow An average  should  be simple  in comprehension  so that it  can he calculated with  reasonable  case and  rapidity else its use will  be  very  limited.

d.It Should  be Based  on all Observation in  the Series An average will be truly representative  of the  whole  mass of data if it is computed from  all the observations.

e It Should  not be Affected Much  by a Few  Extreme Values A few very  small  or very large  observations should  not unduly affect  the value  of a good average.

f. It Should be Capable of Further Algebraic Treatment An average  should lend itself readily to  further algebraic treatment. For example  if average of two  or more  sets   of data  are  known  it  should be  possible  to obtain  the average of combined group.

g.It Should  be Capable of Being  Used  in Further Statistical Computation or Processing's An average  should  posses  this  quality e, g,  arithmetic  mean is  suitable for  calculating standard  deviation  else it will not be  of great  use in  further statistical   analysis and its utility will be limited.

It Should  possess Sampling  Stability An average  should  be least affected by sampling. By  this  we mean  that if  we take  independent  random  samples  of the same  size from  a given  population and  compute  the average   each  of these samples in an ideal average, the  values  so obtained  from different  samples should  not vary  much  from  one  another