From 46632b42fb438dd55d0e9da9f5ca13614cd771aa Mon Sep 17 00:00:00 2001
From: Marko Lalovic <ml.markolalovic@gmail.com>
Date: Fri, 21 Jun 2024 11:29:39 +0200
Subject: [PATCH] response_prop added to tests

---
 dev/dev.R                        |  1 -
 man/simulate_likert.Rd           |  6 +++---
 tests/testthat/test_simulation.R | 12 +++++-------
 3 files changed, 8 insertions(+), 11 deletions(-)

diff --git a/dev/dev.R b/dev/dev.R
index 16c3210..7d62418 100644
--- a/dev/dev.R
+++ b/dev/dev.R
@@ -1,3 +1,2 @@
 # Development
 devtools::load_all("/Users/markolalovic/dev/latent2likert/")
-
diff --git a/man/simulate_likert.Rd b/man/simulate_likert.Rd
index e6d5613..daa4ad8 100644
--- a/man/simulate_likert.Rd
+++ b/man/simulate_likert.Rd
@@ -21,7 +21,7 @@ Simulates Likert scale item responses based on a specified number
 of response categories and the centered parameters of the latent variable.
 }
 \details{
-The simulation process uses the following model proposed by Boari and Nai-Ruscone.
+The simulation process uses the following model detailed by Boari and Nai-Ruscone.
 Let \eqn{X} be the continuous variable of interest, measured using Likert scale
 questions with \eqn{K} response categories. The observed discrete variable \eqn{Y} is
 defined as follows:
@@ -58,8 +58,8 @@ simulate_likert(n_levels = 7, cp = cp2)
 }
 \references{
 Boari, G. and Nai Ruscone, M. (2015).
-A procedure simulating likert scale item responses.
-\emph{Electronic journal of applied statistical analysis} \bold{8(3)}, 288–297.
+A procedure simulating Likert scale item responses.
+\emph{Electronic Journal of Applied Statistical Analysis} \bold{8(3)}, 288–297.
 \doi{10.1285/i20705948v8n3p288}
 }
 \seealso{
diff --git a/tests/testthat/test_simulation.R b/tests/testthat/test_simulation.R
index cb3ae1b..03c16ce 100644
--- a/tests/testthat/test_simulation.R
+++ b/tests/testthat/test_simulation.R
@@ -5,7 +5,7 @@ testthat::test_that("proportions of generated responses match actual probabiliti
     n_levels <- 5
     n_items <- 3
 
-    mean <- c(0, -1, -1)
+    mean <- c(-1, 0, 1)
     sd <- c(1, 1, 0.5)
     skew <- c(0.5, 0.5, 0.5)
     corr <- 0.5
@@ -13,9 +13,8 @@ testthat::test_that("proportions of generated responses match actual probabiliti
     data <- rlikert(size, n_items, n_levels, mean, sd, skew, corr)
     for (i in seq_len(n_items)) {
         cp <- c("mu"=mean[i], "sd"=sd[i], "skew"=skew[i])
-        prob <- simulate_likert(5, cp)
-        prob <- as.numeric(prob)
-        data_prop <- as.numeric(prop.table(table(data[,i])))
+        prob <- simulate_likert(n_levels, cp)
+        data_prop <- response_prop(data[,i], n_levels)
         testthat::expect_equal(prob, data_prop, tolerance = 0.05)
     }
 })
@@ -34,9 +33,8 @@ testthat::test_that("proportions of generated responses match actual probabiliti
     data <- rlikert(size, n_items, n_levels, mean, sd, skew, corr)
     for (i in seq_len(n_items)) {
         cp <- c("mu"=mean[i], "sd"=sd[i], "skew"=skew[i])
-        prob <- simulate_likert(5, cp)
-        prob <- as.numeric(prob)        
-        data_prop <- as.numeric(prop.table(table(data[,i])))
+        prob <- simulate_likert(n_levels, cp)
+        data_prop <- response_prop(data[,i], n_levels)
         testthat::expect_equal(prob, data_prop, tolerance = 0.05)
     }
 })