periphery on the other. If it would represent agglomeration effects, this would mean that
its center of gravity has shifted southward.
The second component seems to be new. It expresses an opposition between the
southern provinces of North Brabant and Limburg, on the one hand, and the central and
northern Netherlands on the other. The lowest loadings for this component are found in
an area that includes Amsterdam, Utrecht, and Almere, with Hilversum as its central
point. It is not easy to interpret this opposition in terms of location factors.
Rotation of three components has yielded results that lend itself rather well to
interpretation, not only for the first two surveys in the Netherlands, but also for similar
surveys that were held in the northern Netherlands and Germany (Meester 2004). The
same is true for Holvoet’s research in Belgium. Thus, the results for the Netherlands in
2003 seem to deviate rather strongly on this point.
Again, it is Dirkzwager’s model that helps to gain more insight in the matter. The
choice to rotate three components for the survey of 2003 was made for reasons of
consistency. Applying the method of Dirkzwager to the data for 2003, it turns out that in
this case, rotation of four components uncovers patterns that are easier to interpret.
Figure 5 shows the loading patterns for these components. It is interesting to note that, if
we rotate four instead of three components for 1983 and 1993 as well, the results prove
to be basically the same for all three surveys. Apparently, the identity of these four
components has not changed during the period under consideration.
Component 1 and 4 look familiar. The first component expresses the opposition
between center and periphery, and can be interpreted as ‘relative location with respect to
the national market’ (Figure 5a). Its center of gravity is close to the city of Utrecht.
Component 4 represents the opposition between the densely populated Randstad area,
and therefore, interpretation as ‘agglomeration effects’ is appropriate (Figure 5d). The
loading patterns of these two components resemble the ones shown in Figure 4a and 4c,
respectively.
The pattern that is displayed by the third component (Figure 5c), shows a number of
characteristics that are also found in Figure 4b. Particularly, the high loadings in the
province of Gelderland and the low loadings in the western Randstad should be
mentioned. Since Gelderland is one of the most highly rated residential areas in the
Netherlands, there might be a relation with residential attractivity as a location factor
here, but the pattern is different in some respects.
11