First functions for potential wald test calculation

- vcov.boot_seminr_model()
- coef.boot_seminr_model()

R/wald.R

@@ -0,0 +1,69 @@
+vcov.boot_seminr_model <- function(boot_obj, endogenous = NULL) {
+  sm <- boot_obj$smMatrix
+  if (!is.null(endogenous)) {
+    sm <- sm[sm[, "target"] %in% endogenous, ]
+  }
+
+  boot_paths <- boot_obj$boot_paths
+
+  boot_path_matrix <- apply(
+    boot_paths,
+    MARGIN = 3,
+    FUN = \(boot_i) apply(sm, MARGIN = 1, FUN = \(path) boot_i[path['source'], path['target']])
+  ) |> t()
+
+  colnames(boot_path_matrix) <- apply(sm, MARGIN=1, \(path) paste(path['source'], path['target'], sep = '->'))
+  cov(boot_path_matrix)
+}
+
+coef.boot_seminr_model <- function(boot_obj, endogenous = NULL) {
+  sm <- boot_obj$smMatrix
+  if (!is.null(endogenous)) {
+    sm <- sm[sm[, "target"] %in% endogenous, ]
+  } else {
+    endogenous <- unique(sm[, "target"])
+  }
+
+  exogenous <- unique(sm[, "source"])
+
+  boot_obj$path_coef[exogenous, endogenous]
+}
+
+# library(aod)
+
+# lm_model <- lm(CUSL1 ~ IMAG1 + CUSA1 + CUSCO, data=mobi)
+# wald.test(b=coef(lm_model), Sigma=vcov(lm_model), Terms=c(2,3))
+
+
+# library(seminr)
+
+# mobi_mm <- constructs(
+#   composite("IMG",        multi_items("IMAG", 1:5)),
+#   composite("EXP",  multi_items("CUEX", 1:3)),
+#   composite("QLT",      multi_items("PERQ", 1:7)),
+#   composite("VAL",        multi_items("PERV", 1:2)),
+#   composite("SAT", multi_items("CUSA", 1:3)),
+#   composite("COM",   single_item("CUSCO")),
+#   composite("LOY",      multi_items("CUSL", 1:3))
+# )
+
+# mobi_sm <- relationships(
+#   paths(from = "IMG", to = c("EXP", "SAT", "LOY")),
+#   paths(from = "EXP", to = c("QLT", "VAL", "SAT")),
+#   paths(from = "QLT", to = c("VAL", "SAT")),
+#   paths(from = "VAL", to = c("SAT")),
+#   paths(from = "SAT", to = c("COM", "LOY")),
+#   paths(from = "COM", to = "LOY")
+# )
+
+# mobi_pls <- estimate_pls(
+#   data = mobi,
+#   measurement_model = mobi_mm,
+#   structural_model = mobi_sm
+# )
+
+# boot_pls <- bootstrap_model(seminr_model = mobi_pls, nboot = 1000)
+
+# b <- coef.boot_seminr_model(boot_obj = boot_pls, endogenous = "LOY")
+# S <- vcov.boot_seminr_model(boot_obj = boot_pls, endogenous = "LOY")
+# wald.test(b=b, Sigma=S, Terms=c(2,3))