Because the selection of a sampling scheme can have strong effects on the a
dmissibility of log-linear models, the authors discuss these effects for th
e multinomial and the product multinomial sampling schemes for prospective
and retrospective sampling. In multinomial sampling, marginal frequencies a
re not fixed a priori. In product multinomial sampling, uni- or multidimens
ional marginal frequencies are considered fixed. Therefore, models must fit
these marginal frequencies exactly. A data example illustrates model selec
tion using data from a project of family development.