2372 intergenerational mobility: core model of social fluidity

We can at least maintain that . . . the . . . effects
that our model comprises do tend to operate
cross nationally in the way anticipated, even
if with differing strengths (Erikson and
Goldthorpe 1992: 174). Nevertheless, even
allowing the parameters of the core model to
differ between countries, they still find that
everywhere except England and France, some
adjustments have to be made to the core model
itself in order successfully to capture national
variations in fluidity. These are sometimes called
the national variants of the core model. The
adjustments are of two kinds. In some cases the
existing matrices are redefined: so, for example,
when applied to the West German data, IN2
assigns immobility in class III to level 1, rather
than 2. That is, there is no tendency for class
inheritance to be higher here than in any other
class. In other cases, additional affinity matrices
were included. So, in Hungary, an additional
matrix, called AFX, captures the tendency for
movement from the class of farmers, IVc, to the
class of agricultural workers, VIIb.

CONCLUSIONS

The core model is nowadays very widely used
(see, for example, the essays in Breen 2004) and
its fit to empirical has often been adequate, and
in some cases, good. The main reason for this is
that it succeeds in capturing the features evident
in most mobility tables, namely the tendency for
clustering on some cells of the main diagonal of
the table (inheritance effects), and the inequality
that derives from a hierarchical ordering of the
classes. But these features are common to almost
all models of fluidity: indeed, given a number
of mobility tables, it is usually quite difficult to
decide which of a set of models is to be preferred
because none provides an unequivocal best fit to
all the tables. If there can be said to be any core
features of fluidity it is the tendency for cluster
ing on the diagonal and the hierarchical compo
nent. The latter can be modeled in various
ways, and, indeed, Erikson and Goldthorpes
approach here is idiosyncratic, with most ana
lysts preferring a simple scaling of the origins
and destinations according to the mean prestige
or mean income of the occupations that make up
each class. Nevertheless, a model which only fits
inheritance and hierarchical effects is unlikely
to fit any mobility table. The claims of the core
model then rest on having found other processes
(instantiated in the se and af terms) that are
common to all fluidity regimes. Unfortunately,
the need to introduce national variants of the
model, which mainly modify the se and af terms,
somewhat undermines that claim. As Sorenson
(1992: 309) remarks: any mobility regime can
be represented as some core model with national
variants.
The major weakness of the core model is that
the effects hypothesized to shape fluidity are all
operationalized as binary contrasts or dummy
variables (on the other hand, this may also help
to account for the models popularity, since one
requires no more than the mobility table itself in
order to employ it). The alternative is to express
log odds ratios as proportional to differences in
scores on measured variables, and several other
models of fluidity adopt this approach (e.g.,
Hout 1984; Breen & Whelan 1992; see also
Breen & Whelan 1994). To illustrate: we might
express the log odds ratios capturing hierarchical
inequality as proportional to the difference in the
mean earnings of the origin and destination
classes involved in the comparison. Clearly, the
choice of explanatory variables should follow
from some hypothesized mechanisms rooted
in individual action and interaction, and, in
fact, the use of explanatory variables inevitably
pushes mobility research away from a concen
tration on tabular data and towards analyses
using data on individuals (Logan 1983; Breen
1994) which test hypotheses about the processes
that shape individual mobility trajectories and
account for the variation among them. A move
in this direction would do much to advance the
study of social fluidity both theoretically and
empirically.

SEE ALSO: Class, Status, and Power; Inter
generational Mobility: Methods of Analysis;
Log Linear Models; Mobility, Horizontal and
Vertical; Mobility, Intergenerational and Intra
generational

REFERENCES AND SUGGESTED
READINGS

Breen, R. (1994) Individual Level Models for Mobi-
lity Tables and Other Cross-Classifications. Socio
logical Methods and Research 23: 147 73.