4) 
- 428 - 
Footnotes 
+ 
I would like to thank Iain Paterson for nelpful com- 
ments. 
1) 
See Sorensen and Sorensen (1985) for an application of 
this model to cohort shifts in age at marriage. 
I would like to thank Gilg Seeber who estimated the pa- 
2) 
rameters of the log-logistic model with a GLIM-Macro. 
Tne good fit of diffusion models does not exclude ai- 
3) 
ternative explanations of the risk pattern. Assumptions 
of heterogeneity, search-models, and a combination of 
search theory, latent states and heterogeneity (Coale 
and McNeil, 1972) offer rival explanations of the ob¬ 
served patterns (see Diekmann, 1986b for a more exten¬ 
sive discussion). 
Tests of the proportional hazards model are discussed 
in Diekmann (1986b). Kaplan-Meier estimates of the sur- 
vival function fit quite well with the model prediction 
for mean values of covariates.