Invariance Partial Pruning Test

An implementation of the Invariance Partial Pruning (IVPP) approach described in Du, X., Johnson, S. U., Epskamp, S. (2025) The Invariance Partial Pruning Approach to The Network Comparison in Longitudinal Data. IVPP is a two-step method that first test for global network structural difference with invariance test and then inspect specific edge difference with partial pruning. The package also allows you to compute centrality measures and use radar chart to plot. Analysis of bridge centralities by community pairs is also possible (e.g., the bridge strength from depression to anxiety, and from depression to panic disorder).

IVPP

An implementation of the Invariance Partial Pruning (IVPP) approach described in Du, X., Johnson, S. U., Epskamp, S. (in prep) to comparing idiographic and panel network models. IVPP is a two-step method that first test for global network structural difference with invariance test and then inspect specific edge difference with partial pruning.

Installation

To install from CRAN:

install.packages("IVPP")

You can install the development version of IVPP from GitHub with:

# install.packages("devtools")
devtools::install_github("xinkaidupsy/IVPP")

Example

An example that uses IVPP to compare panelGVAR models:

library(IVPP)

# Generate the network
net_ls <- gen_panelGVAR(n_node = 6,
                        p_rewire_temp = 0.5,
                        p_rewire_cont = 0.5,
                        n_group = 2)

# Generate the data
data <- sim_panelGVAR(temp_base_ls = net_ls$temporal,
                      cont_base_ls = net_ls$omega_zeta_within,
                      n_person = 200,
                      n_time = 3,
                      n_group = 2,
                      n_node = 6)

# global test on both nets
omnibus_both <- IVPP_panelgvar(data,
                               vars = paste0("V",1:6),
                               idvar = "subject",
                               beepvar = "time",
                               groups = "group",
                               g_test_net = "both",
                               net_type = "sparse",
                               partial_prune = FALSE,
                               ncores = 2)

# global test on temporal
omnibus_temp <- IVPP_panelgvar(data,
                               vars = paste0("V",1:6),
                               idvar = "subject",
                               beepvar = "time",
                               groups = "group",
                               g_test_net = "temporal",
                               net_type = "sparse",
                               partial_prune = FALSE,
                               ncores = 2)

# global test on cont
omnibus_cont <- IVPP_panelgvar(data,
                               vars = paste0("V",1:6),
                               idvar = "subject",
                               beepvar = "time",
                               groups = "group",
                               g_test_net = "contemporaneous",
                               net_type = "sparse",
                               partial_prune = FALSE,
                               ncores = 2)

# partial prune on both networks
pp_both <- IVPP_panelgvar(data,
                          vars = paste0("V",1:6),
                          idvar = "subject",
                          beepvar = "time",
                          groups = "group",
                          global = FALSE,
                          partial_prune = TRUE,
                          prune_net = "both",
                          ncores = 2)

An example that uses IVPP to compare N = 1 GVAR models

library(IVPP)

# Generate the network
net_ls <- gen_tsGVAR(n_node = 6,
                     p_rewire_temp = 0.5,
                     p_rewire_cont = 0.5,
                     n_persons = 2)

# Generate the data
data <- sim_tsGVAR(beta_base_ls = net_ls$beta,
                   kappa_base_ls = net_ls$kappa,
                   # n_person = 2,
                   n_time = 300)
                   
# global test on both networks
omnibus_both <- IVPP_tsgvar(data,
                            vars = paste0("V",1:6),
                            idvar = "id",
                            g_test_net = "both",
                            net_type = "sparse",
                            partial_prune = FALSE,
                            ncores = 2)
                            
# global test on temporal
omnibus_temp <- IVPP_tsgvar(data,
                            vars = paste0("V",1:6),
                            idvar = "id",
                            g_test_net = "temporal",
                            net_type = "sparse",
                            partial_prune = FALSE,
                            ncores = 2)

# global test on cont
omnibus_cont <- IVPP_tsgvar(data,
                            vars = paste0("V",1:6),
                            idvar = "id",
                            g_test_net = "contemporaneous",
                            net_type = "sparse",
                            partial_prune = FALSE,
                            ncores = 2)

# partial prune on both networks
pp_both <- IVPP_tsgvar(data,
                       vars = paste0("V",1:6),
                       idvar = "id",
                       global = FALSE,
                       partial_prune = TRUE,
                       prune_net = "both",
                       ncores = 2)