Skip to contents

Simulate network data from a Social Relations Model

Source: R/sim_srm.R
sim_srm.Rd

Generates synthetic network matrices from known actor, partner, and unique effects, useful for teaching, validation, and power analysis.

Usage

sim_srm(
  n_actors = 10,
  n_time = 1,
  actor_var = 1,
  partner_var = 1,
  unique_var = 1,
  actor_partner_cov = 0,
  relationship_cov = 0,
  grand_mean = 0,
  bipartite = FALSE,
  seed = NULL
)

Arguments

n_actors

Number of actors (nodes) in the network. For bipartite networks, a length-2 vector c(n_row, n_col) specifying the number of senders and receivers.

n_time

Number of time periods (default 1 for cross-sectional).

actor_var

Variance of actor (sender) effects.

partner_var

Variance of partner (receiver) effects.

unique_var

Variance of unique dyadic effects.

actor_partner_cov

Covariance between actor and partner effects. Must satisfy actor_partner_cov^2 <= actor_var * partner_var.

relationship_cov

Covariance of unique effects within dyads (generalized reciprocity). Must satisfy abs(relationship_cov) <= unique_var.

grand_mean

Overall network mean.

bipartite

If TRUE, generate a bipartite (two-mode) network. When bipartite, n_actors is a length-2 vector c(n_row, n_col). The actor_partner_cov and relationship_cov parameters are ignored for bipartite networks (these quantities are not defined when senders and receivers are from different populations).

seed

Optional random seed for reproducibility.

Value

A list with components:

Y

A matrix or named list of matrices (if n_time > 1).

truth

A list containing the true actor_effects, partner_effects, unique_effects, and grand_mean.

params

The parameter values used for simulation.

Examples

# cross-sectional
sim = sim_srm(n_actors = 10, actor_var = 1, partner_var = 0.5,
               unique_var = 2, seed = 6886)
fit = srm(sim$Y)
summary(fit)
#> Social Relations Model - Variance Decomposition
#> ================================================== 
#> 
#> Component                Variance  % Total
#> ------------------------------------------ 
#> Actor                      0.4843    15.2%
#> Partner                    0.6512    20.4%
#> Unique                     2.0533    64.4%
#> Relationship (cov)        -0.1958       --
#> Actor-Partner (cov)        0.5433       --

# \donttest{
# longitudinal
sim_long = sim_srm(n_actors = 8, n_time = 5, seed = 6886)
fit_long = srm(sim_long$Y)
fit_long
#> Social Relations Model
#> ---------------------------------------- 
#> Mode:       unipartite
#> Time points: 5
#> Actors:      8
#> Grand mean:  -0.4985 (avg)
#> 
#> Variance Decomposition:
#>   Actor          0.8418  ( 29.1%)
#>   Partner        1.0098  ( 34.9%)
#>   Unique         1.0388  ( 35.9%)
#>   Relationship  -0.0524  (cov)
#>   Actor-Partner   0.0750  (cov)

# bipartite
sim_bip = sim_srm(n_actors = c(6, 8), bipartite = TRUE, seed = 6886)
fit_bip = srm(sim_bip$Y)
summary(fit_bip)
#> Social Relations Model - Variance Decomposition
#> ================================================== 
#> 
#> Component                Variance  % Total
#> ------------------------------------------ 
#> Actor                      1.2467    39.3%
#> Partner                    0.8214    25.9%
#> Unique                     1.1075    34.9%
# }