Skip to contents

Thin wrapper around Nestimate::centrality_stability() that validates the input is an htna network (or htna_group), forwards the call, and tags the result with an htna_stability class so callers can dispatch on the htna form.

Usage

centrality_stability_htna(
  x,
  measures = c("InStrength", "OutStrength", "Betweenness"),
  iter = 1000L,
  drop_prop = seq(0.1, 0.9, by = 0.1),
  threshold = 0.7,
  certainty = 0.95,
  method = "pearson",
  centrality_fn = NULL,
  loops = FALSE,
  seed = NULL
)

Arguments

x

An htna network from build_htna() or an htna_group.

measures

Character vector of centrality measures to assess. Default c("InStrength", "OutStrength", "Betweenness"): the three measures whose values are bit-equal between htna's cograph engine and Nestimate's default centrality implementation. To assess htna's nine measures (including the three that differ between engines: BetweennessRSP, Diffusion, Clustering), pass them explicitly together with a custom centrality_fn.

iter

Integer. Number of resamples per drop proportion. Default 1000.

drop_prop

Numeric vector. Proportions of sessions to drop. Default seq(0.1, 0.9, by = 0.1).

threshold

Numeric in [0, 1]. Correlation threshold for the case-dropping stability coefficient. Default 0.7.

certainty

Numeric in [0, 1]. Probability of meeting the threshold required for a drop proportion to count as stable. Default 0.95.

method

Correlation method, one of "pearson", "spearman", "kendall". Default "pearson".

centrality_fn

Optional function to compute centralities. Default NULL (use Nestimate's internal implementation). See Nestimate::centrality_stability() for the expected signature.

loops

Logical. Whether to retain self-loops. Default FALSE.

seed

Optional integer seed for reproducibility. Default NULL (no seed reset).

Value

For a single htna network, an object of class c("htna_stability", "net_stability") with the same components as Nestimate::centrality_stability(): cs (the stability coefficients, one per measure), correlations (per-drop-proportion correlation traces), and the parameters that were used. summary() returns the tidy table: one row per (measure, drop proportion), with columns measure, drop_prop, mean_cor, sd_cor and prop_above. See print.htna_stability() and plot.htna_stability(). For an htna_group, a named list of such objects with class c("htna_stability_group", "list").

Details

For an htna_group, the call is iterated per cohort and the result is a named list (one htna_stability per cohort), with class c("htna_stability_group", "list").

References

Epskamp, S., Borsboom, D., & Fried, E. I. (2018). Estimating psychological networks and their accuracy: A tutorial paper. Behavior Research Methods, 50(1), 195-212. doi:10.3758/s13428-017-0862-1

Examples

data(human_ai)
net <- build_htna(human_ai, actor_type = "actor_type")
#> Metadata aggregated per session: ties resolved by first occurrence in 'session_date' (1 sessions), 'cluster' (42 sessions), 'actor_type' (24 sessions)
cs  <- centrality_stability_htna(net, iter = 30, seed = 1)
cs            # CS-coefficient per measure
#> HTNA Centrality Stability Analysis
#> ===================================
#> Centrality Stability (30 iterations, threshold = 0.7)
#>   Drop proportions: 0.1, 0.2, 0.3, 0.4, 0.5, 0.6, 0.7, 0.8, 0.9
#> 
#>   CS-coefficients:
#>     InStrength       0.90
#>     OutStrength      0.90
#>     Betweenness      0.80
summary(cs)   # mean correlation per measure and drop proportion
#>        measure drop_prop  mean_cor       sd_cor prop_above
#> 1   InStrength       0.1 0.9997933 9.399327e-05  1.0000000
#> 2   InStrength       0.2 0.9995121 2.589640e-04  1.0000000
#> 3   InStrength       0.3 0.9989968 5.254436e-04  1.0000000
#> 4   InStrength       0.4 0.9987609 7.592632e-04  1.0000000
#> 5   InStrength       0.5 0.9979411 1.335385e-03  1.0000000
#> 6   InStrength       0.6 0.9969654 1.586312e-03  1.0000000
#> 7   InStrength       0.7 0.9960166 2.181399e-03  1.0000000
#> 8   InStrength       0.8 0.9916548 5.620265e-03  1.0000000
#> 9   InStrength       0.9 0.9832595 8.474270e-03  1.0000000
#> 10 OutStrength       0.1 0.9987117 1.133226e-03  1.0000000
#> 11 OutStrength       0.2 0.9974190 2.067790e-03  1.0000000
#> 12 OutStrength       0.3 0.9960323 2.227121e-03  1.0000000
#> 13 OutStrength       0.4 0.9942741 2.601673e-03  1.0000000
#> 14 OutStrength       0.5 0.9860923 1.021467e-02  1.0000000
#> 15 OutStrength       0.6 0.9847606 8.605913e-03  1.0000000
#> 16 OutStrength       0.7 0.9775272 1.271530e-02  1.0000000
#> 17 OutStrength       0.8 0.9553243 2.549594e-02  1.0000000
#> 18 OutStrength       0.9 0.9259479 4.487740e-02  1.0000000
#> 19 Betweenness       0.1 0.9895928 9.718681e-03  1.0000000
#> 20 Betweenness       0.2 0.9786379 1.764171e-02  1.0000000
#> 21 Betweenness       0.3 0.9747873 1.717430e-02  1.0000000
#> 22 Betweenness       0.4 0.9700564 2.368309e-02  1.0000000
#> 23 Betweenness       0.5 0.9576797 3.076986e-02  1.0000000
#> 24 Betweenness       0.6 0.9564011 2.816820e-02  1.0000000
#> 25 Betweenness       0.7 0.9406976 3.667811e-02  1.0000000
#> 26 Betweenness       0.8 0.9328796 4.775810e-02  1.0000000
#> 27 Betweenness       0.9 0.8714616 7.999922e-02  0.9333333