Merge pull request #144 from sfcheung/devel

0.1.14.10: Change the default of `use_implied_stats` to `TRUE` and a note will be printed in the plot

DESCRIPTION

@@ -1,6 +1,6 @@
 Package: manymome
 Title: Mediation, Moderation and Moderated-Mediation After Model Fitting
-Version: 0.1.14.9
+Version: 0.1.14.10
 Authors@R:
     c(person(given = "Shu Fai",
            family = "Cheung",

NEWS.md

@@ -1,4 +1,4 @@
-# manymome 0.1.14.9
+# manymome 0.1.14.10
 
 ## New Features
 
@@ -41,11 +41,12 @@
   makes more sense for multigroup
   models, in which the distribution of
   variables are allowed to be different
-  between groups. If desired, users
-  can use the model implied statistics
-  to determine the means and SDs, which
+  between groups. Since 0.1.14.10,
+  by default, the model implied statistics
+  are used to determine the means and SDs
+  used in drawing he plot. This approach
   is useful when equality constraints
-  are present. (0.1.14.2, 0.1.14.6)
+  are present. (0.1.14.2, 0.1.14.6, 0.1.14.10)
 
 - The `plot`-method of
   `cond_indirect_effects`-class objects

R/plotmod.R

@@ -161,11 +161,12 @@
 #'
 #' @param use_implied_stats For a
 #' multigroup model, if `TRUE`,
+#' the default,
 #' model implied statistics will be
 #' used in computing the means and SDs,
 #' which take into equality constraints,
 #' if any.
-#' If `FALSE`, the default, then the raw
+#' If `FALSE`, then the raw
 #' data is
 #' used to compute the means and SDs.
 #' For latent variables, model implied
@@ -263,7 +264,7 @@ plot.cond_indirect_effects <- function(
                             line_width = 1,
                             point_size = 5,
                             graph_type = c("default", "tumble"),
-                            use_implied_stats = FALSE,
+                            use_implied_stats = TRUE,
                             ...
                     ) {
     has_groups <- cond_indirect_effects_has_groups(x)
@@ -298,6 +299,7 @@ plot.cond_indirect_effects <- function(
         stop("x_method cannot be 'percentile' if x is a latent variable.")
       }
     graph_type <- match.arg(graph_type)
+    graph_type_original <- graph_type
     if (has_groups && graph_type == "default") {
         # warning("Only tumble graph is supported for multiple group models. ",
         #         "Changed graph_type to 'tumble'.")
@@ -586,11 +588,16 @@ plot.cond_indirect_effects <- function(
             cap_txt <- cap_std
           }
       }
-    if (has_groups && graph_type == "default") {
+    if (has_groups && graph_type_original == "default") {
         cap_txt <- paste0(cap_txt,
                           "\n",
                           "Graph type is set to tumble for multiple group models.")
       }
+    if (use_implied_stats && fit_type != "lm") {
+        cap_txt <- paste0(cap_txt,
+                          "\n",
+                          "Model implied means and SDs are used in drawing the points'.")
+      }
     out <- p +
       ggplot2::labs(title = title,
                     caption = cap_txt) +

README.md

@@ -9,7 +9,7 @@
 [![R-CMD-check](https://github.com/sfcheung/manymome/actions/workflows/R-CMD-check.yaml/badge.svg)](https://github.com/sfcheung/manymome/actions/workflows/R-CMD-check.yaml)
 <!-- badges: end -->
 
-(Version 0.1.14.9, updated on 2024-04-01, [release history](https://sfcheung.github.io/manymome/news/index.html))
+(Version 0.1.14.10, updated on 2024-04-01, [release history](https://sfcheung.github.io/manymome/news/index.html))
 
 # manymome  <img src="man/figures/logo.png" align="right" height="150" />
 

rebuild_vignettes.R

@@ -18,6 +18,6 @@ setwd(base_dir)
 base_dir <- getwd()
 
 setwd("vignettes/articles")
-knitr::knit("med_mg.Rmd.originaL", output = "med_mg.Rmd")
+knitr::knit("med_mg.Rmd.original", output = "med_mg.Rmd")
 
 setwd(base_dir)

tests/testthat/test_mg_boot.R

@@ -829,10 +829,10 @@ tmp3 <- cond_indirect_effects(x = "x",
                               y = "m1",
                               fit = fit_med_meq3)
 
-p1a <- plot(tmp1)
-p2a <- plot(tmp2)
+p1a <- plot(tmp1, use_implied_stats = FALSE)
+p2a <- plot(tmp2, use_implied_stats = FALSE)
 p2b <- plot(tmp2, use_implied_stats = TRUE)
-p3a <- plot(tmp3)
+p3a <- plot(tmp3, use_implied_stats = FALSE)
 p3b <- plot(tmp3, use_implied_stats = TRUE)
 
 test_that("plot.cond_indirect_effects: multiple groups, implied stats", {

vignettes/articles/med_mg.Rmd

@@ -1,7 +1,7 @@
 ---
 title: "Multigroup Models With Mediation Effects"
 author: "Shu Fai Cheung & Sing-Hang Cheung"
-date: "2024-03-31"
+date: "2024-04-03"
 output:
   html_document:
     fig.align: "center"
@@ -204,7 +204,7 @@ fit_boot_out <- do_boot(fit = fit,
                         seed = 53253,
                         ncores = 8)
 #> 8 processes started to run bootstrapping.
-#> The expected CPU time is about 0 second(s).
+#> The expected CPU time is about 7.5 second(s).
 ```
 
 # Estimate Indirect Effects
@@ -889,20 +889,37 @@ for multigroup models:
 Note that, for multigroup models, the *tumble*
 graph proposed by @bodner_tumble_2016 will
 always be used. The position of a line for
-a group is determined by the descriptive statistics
-of this group (e.g, means and SDs). For example,
+a group is determined by the model implied
+means and SDs of this group. If no equality
+constraints imposed, these means and SDs are
+close to the sample means and SDs.
+For example,
 the line segment of `"Group A"` is far to the right
 because `"Group A"` has a larger mean of `m` than
-`"Group B"`:
+`"Group B"`.
+
+These are the model implied means and SDs:
 
 
 ```r
-by(dat$m, dat$group, mean)
-#> dat$group: Group A
-#> [1] 18.43357
-#> ------------------------------------------------------------ 
-#> dat$group: Group B
-#> [1] 13.42327
+# Model implied means
+lavInspect(fit, "mean.ov")
+#> $`Group A`
+#>      m      y      x     c1     c2 
+#> 18.434 17.973  9.985  2.055  4.883 
+#> 
+#> $`Group B`
+#>      m      y      x     c1     c2 
+#> 13.423 18.915 10.046  2.138  5.088
+
+# Model implied SDs
+tmp <- lavInspect(fit, "cov.ov")
+sqrt(diag(tmp[["Group A"]]))
+#>         m         y         x        c1        c2 
+#> 1.4916076 2.2158085 1.1055723 0.9687943 1.0741628
+sqrt(diag(tmp[["Group B"]]))
+#>         m         y         x        c1        c2 
+#> 1.2142143 2.5838604 1.0094330 0.9518064 1.0594076
 ```
 
 It would be misleading if the two lines are plotted on the

vignettes/articles/med_mg.Rmd.original

@@ -522,14 +522,25 @@ plot(path1)
 Note that, for multigroup models, the *tumble*
 graph proposed by @bodner_tumble_2016 will
 always be used. The position of a line for
-a group is determined by the descriptive statistics
-of this group (e.g, means and SDs). For example,
+a group is determined by the model implied
+means and SDs of this group. If no equality
+constraints imposed, these means and SDs are
+close to the sample means and SDs.
+For example,
 the line segment of `"Group A"` is far to the right
 because `"Group A"` has a larger mean of `m` than
-`"Group B"`:
+`"Group B"`.
+
+These are the model implied means and SDs:
 
 ```{r}
-by(dat$m, dat$group, mean)
+# Model implied means
+lavInspect(fit, "mean.ov")
+
+# Model implied SDs
+tmp <- lavInspect(fit, "cov.ov")
+sqrt(diag(tmp[["Group A"]]))
+sqrt(diag(tmp[["Group B"]]))
 ```
 
 It would be misleading if the two lines are plotted on the