QAP: MRQAP for unnested data

In timonelmer/netglm: netglm - generalized linear models for network data

MRQAP for unnested data

Description

Estimates a Multiple Regression Quadratic Assignment Proccedure model (MRQAP; Krackhardt, 1988). MRQAPs allow investigating associations between characteristics of dyads in networks (e.g., the level of homophily between two actors) and a binary or continuous tie variable (e.g., friendship, amount of time spent together).

Usage

QAP(dv, iv1, iv.names, mode = "yQAP", samples = 1000, diag = F, directed = F)

Arguments

dv	a matrix with n * m dimensions and cells indicating the presence of a tie (1 = tie, 0 = no tie for binary variables) or the weight of a tie (for continous tie variables) charcterizing the dependent variable
iv1	a list of matrices with n * m dimensions characterizing the independent variables
iv.names	names of the independent variables for the output object
mode	permutation method to be applied. default is "yQAP" for permuting the Y / dv variables."dspQAP" applies Dekker's semi partialing method (Dekker, Krackhard, & Snijders, 2007)
samples	number of permutations, default is 1000.
diag	boolean for using the diagonal values of matrices in the estimation. default is FALSE
directed	"directed" if the dependent network is directed (ties from A to B and B to A are possible), "undirected" if the dependent network is undirected (ties from A to B are identical to B to A). Default is "directed".
iv.list.per	lists in the ivs argument are should be nested by group and independent matrices, if this is not the case (grouped by independent matrices and then groups) the argument iv.list.per = "iv" can be used to restructure the data.
family	family of the generalized linear model. default is "gaussian" for continuous dependent varaibles. F or binday dependent variables "binomial" is advised.
round.to	numeric, numer of digits in output table
cpu	number of cpu's to be used for estimation, default is 1
logfilename	name of log file printing intermediate reports during the estimation procedure.
verbose	reports of what is happening under the hood during the call of the function, default is TRUE
global.deltas	during "dspQAP" estimation, should global or local delta values be used. default is TRUE
return.perms	should permuted networks be part of the output? default is FALSE

References

Dekker, D.; Krackhardt, D.; Snijders, T.A.B. (2007). “Sensitivity of MRQAP Tests to Collinearity and Autocorrelation Conditions.” Psychometrika, 72(4), 563-581.

Krackhardt, D. (1987). “QAP Partialling as a Test of Spuriousness.” Social Networks, 9 171-186.

Krackhardt, D. (1988). “Predicting With Networks: Nonparametric Multiple Regression Analyses of Dyadic Data.” Social Networks, 10, 359-382.

See Also

QAP.MG

Examples

# create test data #
inspired by the example funciton in sna::netlm
ivnet1<-sna::rgraph(20,4)

dv1<-ivnet1[1,,]+4*ivnet1[2,,]+2*ivnet1[3,,]   # Note that the fourth graph is unrelated
dv1 <- dv1 + rnorm(400,mean = 1, sd = 1)
iv1 <- list(ivnet1[1,,],ivnet1[2,,],ivnet1[3,,], ivnet1[4,,])
 QAP.MG(list(dv1), list(iv1), iv.names = c("intercept",paste0("IV",1:4)), samples = 3000)

timonelmer/netglm documentation built on Aug. 14, 2024, 9:39 p.m.