The data is available at the level of individual ages. Age is an important factor in sexual risk behaviour, explaining much of the variance. The age groups used currently (15-19, 20-24 and 25-29) are relatively large, and may encompass a large amount of heterogeneity, particularly for the younger age category. This improvement would model at the level of individual ages. More structure may be wise for the age random effects, such as a random walk, to smooth out excess variation.
The first step would be to make the data available at the level of individual years. The next step is to change the modelling scripts to reflect this change.
The data is available at the level of individual ages. Age is an important factor in sexual risk behaviour, explaining much of the variance. The age groups used currently (15-19, 20-24 and 25-29) are relatively large, and may encompass a large amount of heterogeneity, particularly for the younger age category. This improvement would model at the level of individual ages. More structure may be wise for the age random effects, such as a random walk, to smooth out excess variation.
The first step would be to make the data available at the level of individual years. The next step is to change the modelling scripts to reflect this change.