Skip to contents

Adjusts the edges coefficient in a dynamic network model simulated in netsim() to preserve the mean degree of nodes in the network. Requires at >= 2. Maintains the num(.g2) epi fields (initialized in sim_nets_t1()) for computing the coefficient adjustment.

Usage

edges_correct(dat, at)

Arguments

dat

Main netsim_dat object containing a networkDynamic object and other initialization information passed from netsim().

at

Current time step.

Value

The updated netsim_dat main list object.

Details

By default the correction counts every active node. That is correct when every active node can form a tie, which is the usual case.

It is not correct when a model carries a subpopulation that stays active but is structurally excluded from the network, for example an age band past a sexual-cessation age whose target statistics are all zero, so that ergm pins its terms off and no tie incident to those nodes can form. The population the correction sees then grows while the population that can hold an edge does not, and the whole of the adjustment lands on the nodes that can. The edges.correct.attr control names a binary nodal attribute identifying the nodes eligible to form ties, and the correction then counts only those.

The size of the error is the size of the excluded share. A model whose excluded band grows to a quarter of the population dilutes mean degree among the remainder by the same quarter, and none of it appears in the network diagnostics, because netdx() runs before any node has been excluded.