However, it is possible to fit interactions between time-varying and time-invariant variables in a fixed effects model.
To allow a better understanding of the effects of deprivation on alcohol consumption, we fitted interactions between the neighbourhood deprivation quintiles and each additional explanatory variable indicating the relevant binary transition.
We tested for interaction by fitting interaction terms between the main variable of interest (Aboriginal or Torres Strait Islander status) and all factors that were found to be statistically significantly associated with food insecurity.
Sex and ethnic differences in growth trajectories were estimated by fitting interaction terms between ethnic and sex groups (four groups: white British boys, Pakistani boys, white British girls and Pakistani girls), the constant term and each of the linear slopes (see the full equation in the supplementary material).
Differences in growth trajectories by maternal ED groups (AN, BN and AN+BN) were estimated by fitting interaction terms in the multilevel models between the maternal ED group and the constant term (representing birth length, PI at birth or BMI at age 2) and each of the polynomial terms.
We could revert to fitting a separate model for each of the four causes of death but for demonstrative purposes we have instead fitted interaction terms between each cause and each of the two variables.
Three models were used for the association study: (1) conventional GWAS (genome-wide association study), fitting a single SNP effect regardless of subspecies origin, (2) interaction GWAS, fitting an interaction between SNP and subspecies-origin, and (3) best variable GWAS, fitting the most significant combination of SNP and sub-species origin.
Complex nonlinear relationships can be fitted, and interactions between predictors handled automatically.
Complex nonlinear relationships can be fit, and interactions between predictors handled automatically.
Fitting an interaction between SNP and subspecies origin resulted in more significant SNPs (i.e. more power) than a conventional GWAS.
