Page 47 - Cultures and Organizations

32 THE CONCEPT OF CULTURE

be made for classifying them as belonging to one type or another. With a
dimensional model, on the contrary, cases can always be scored unambigu-
ously. On the basis of their dimension scores, cases can afterward empiri-
cally be sorted into clusters with similar scores. These clusters then form

an empirical typology. More than fifty countries in the IBM study could,
on the basis of their scores on the four dimensions, be sorted into twelve
such clusters. 5
In practice, typologies and dimensional models are complementary.
Dimensional models are preferable for research, and typologies are use-
ful for teaching purposes. This book will use a kind of typology approach
for explaining each of the dimensions. For every separate dimension, it
describes the two opposite extremes as pure types. Later on, some dimen-
sions are plotted two by two, every plot creating four types. The country
scores on the dimensions will show that most real cases are somewhere in
between the extremes.

Using Correlations

Dimensions are based on correlations. Two measures (called variables) are
said to be correlated if they vary together. For example, if we were to
measure the height and weight of a hundred people randomly picked from

the street, we would find the height and weight measures to be correlated:
taller people would also usually be heavier, and shorter ones would also
tend to be lighter. Because some people are tall and skinny and some are
short and fat, the correlation would not be perfect.
6
The coeffi cient of correlation expresses the strength of the relationship.
If the correlation is perfect, so that one measure follows entirely from the
other, the coefficient takes the value 1.00. If the correlation is nonexis-

tent—the two measures are completely unrelated—the coefficient is 0.00.

The coefficient can become negative if the two measures are each other’s

opposite—for example, a person’s height and the number of times he or she
would meet someone who is still taller. The lowest possible value is 1.00;
in this case the two measures are again perfectly correlated, only the one
is positive when the other is negative, and vice versa. In the example of the
height and weight of people, one could expect a coefficient of about 0.80

if the sample included only adults—and even higher if both children and
adults were included in the sample, because children are extremely small
and light compared with adults.