A discussion of alternative ways of modelling and interpreting mixture data

Bjørkestøl, Kirsten; Næs, Tormod


Utgiver: Høgskolen i Agder

Antall sider: 33

Internasjonale standardnumre:
Trykt: 8271174401

Open Access: none

This paper compares different models for regression where some of the variables are mixture components. The focus is on ideas about how to set up a model that is useful for analysing data which do not necessarily come from an ideal statistical design. Recommendations and cautions about the use of som regression procedures in some statistical software are given.