Page 319 - Becoming Metric Wise
P. 319

311
                                                           The Informetric Laws

                 Finally, we come to the most important requirement, known as
              Dalton’s transfer principle (Dalton, 1920). In terms of monetary values
              this requirement states that if one takes money from a poorer person and
              gives it to a richer one (a so-called elementary transfer) then the concen-
              tration of richness increases. Mathematically this requirement can be
              expressed as follows:
                 If x i # x j and 0 , k # x i then

                 Fx 1 ; .. .; x i ; .. .; x j ; .. .; x N , Fx 1 ; .. .; x i 2 k; .. .; x j 1 k; .. .; x N
                                                                         (9.22)
                 The transformation from the left-hand side of (9.22) to the right-hand
              side is called an elementary transfer. The fact that the Lorenz curve is the
              correct tool to describe concentration follows from the following
              theorem.

              Theorem
              (Muirhead, 1903): Let X 5 (x 1 ,.. .,x N ) and Y 5 (y 1 ,.. .,y N ) be decreasing
              arrays and let L(X) and L(Y) be their corresponding Lorenz curves. If
              L(X) is situated above L(Y) (but not coinciding) then X can be generated
              from Y by a finite number of elementary transfers. The reverse is also
              true.
                 Besides the Gini index (see Subsection 4.10.3) also the coefficient of
              variation defined as the standard deviation divided by the arithmetic
              mean is a measure which, for fixed N, satisfies all the requirements men-
              tioned above. Another one is the Simpson index, also known as the
              Hirschmann-Herfindahl index. This measure, denoted as λ, is defined as:

                      N           N              N
                     X       2   X              X
                          x i
                                     2
              λðXÞ 5          5     p with M 5      x i the total number of itemsÞ
                                                      ð
                          M          i
                      i51        i51             i51
                                                                         (9.23)
              9.6 MEASURING DIVERSITY FOR A VARIABLE NUMBER OF
              CELLS

              If F is an acceptable measure of concentration which never takes the
              value 0, then G 5 1/F is an acceptable measure of diversity. Similarly, if F
              is an acceptable measure of concentration, bounded by 1, then G 5 1 2 F
              is an acceptable measure of diversity.
   314   315   316   317   318   319   320   321   322   323   324