In these about ten years, some important contributions have been made by log-linear models (LLM) in the context of mobility tables. However, there are some ambiguous aspects in models themselves as well as results proposed by them.One of the most important question is whether log-linear models can really control for structural mobility, which is defined as marginal descrepancies.
In the first place, this paper attempt to show that the question above can be specified as follows :
Independent Model
_↔_Circulation Mobility does not exist
Because in LLM, “Circulation Mobility” should correspond to deviation of the frequency distribution from an independent model. It will be shown that the proposition above is fault.
However, we can understand marginal effects in LLM as “Structural Mobility” under the following two senses.
1) Marginal effects measure the effect of category size and its change.
2) We can understand such effects as a external condition, that is, as a “force” which enforces mobility structurally.
Now the problem should be that what means these two different definitions can propose for the concept of “Structural Mobility”. For example, under these definitions, are structural mobility and circulation mobility regulated by some different principles (i.e. stratification structure) ?
抄録全体を表示