amp.modeller: Model age-mixing pattern.
In wdelva/RSimpactHelp: Pre- and Post-simulation helper functions for RSimpact Cyan

Description Usage Arguments Details Value Examples

Models the age-mixing pattern for the population simulated in Simpact. The user specifies a time point, time window, and age group for which they would like to obtain a summary of the age-mixing pattern. The user also specifies whether only relationships that started within the time window should be used.

1 2	amp.modeller(dataframe, agegroup, timepoint, timewindow, start = FALSE, SHIMS = TRUE, method = "lmer")

`dataframe`	The dataframe that is produced by `agemix.episodes.df.maker`
`agegroup`	Boundaries of the age group that should be retained, e.g. c(15, 30). The interval is closed on the left and open on the right.
`timepoint`	Point in time during the simulation to be used in the calculation.
`timewindow`	The length of time before the timepoint for which relationships should be included, e.g. 1, representing one year before the timepoint. This should be a whole number.
`start`	This is a logical indicating that only relationships starting after the beginning of the window should be used. If start = FALSE relationships could start before the time window. This is the default.
`SHIMS`	Should only the three most recent relationships be kept, like in the SHIMS survey?
`method`	Should the `link[nlme]{lme}` or `link[lme4]{lmer}` function be used? The lme function can be used to model heteroskedastic residual error.

The amp.modeller function can use either the link[nlme]{lme} or the link[lme4]{lmer} function to build a linear mixed effects model regressing the partner's age at the time the relationship started on his/her own age at the beginning of the relationship. The models are stratified by gender. Each person can have more than one relationship so a random intercept at the level of the person is added to the model.

The models built with link[nlme]{lme} also explicitly account for heteroskedastic variance, since the variance in partner ages may grow for older ages.

The function produces a fitted model.

returns a model fitted with link[nlme]{lme} or link[lme4]{lmer}

1
2
3

data(persreldf)
agemixpatdat <- amp.modeller(dataframe = persreldf, agegroup = c(18, 50),
timepoint = 30, timewindow = 0.5, start = FALSE, SHIMS = TRUE, method = "lmer")