Page 50 - Lungshan Pottery Lunshanoid Research 1977 Paper

37

presence and the absence of each variable in that unit.

According to the coded value of the presence and absence

of each variable i n every single data unit, we can decide

the degree of resemblance of any pair of data units.

The degree of resemblance of any pair of data units

can be described by a concrete figure — a similarity coe-

fficient. According to Sokal and Sneath (1963* 129-130),

and Anderberg (1973* 89), there are many formulas to

calculate different similarity coefficients. After figuring

out the similarity coefficients, not only are we able to

do cluster analysis with them, but one can also do multi-

dimensional scaling using these figures. The value of

these coefficients i s always between 0.0 and 1.0. The

higher the value of the similarity coefficients the more

resemblance between the pair of data units. In addition

to the similarity coefficient, the degree of the resemblance

among the data units also can be described by the distance

of the difference among the data units. The distance of

the difference of data units is calculated by the formula

of 1.0-coefficient. Accordingly, the lower the value of

the distance the higher the degree of the resemblance

among data units.