Add notes and simcode

.gitignore

@@ -1,2 +1,3 @@
 20240918.pdf
 deenadayalan_et_al_2008.pdf
+Rplots.pdf

20241016.R

@@ -0,0 +1,72 @@
+#!/bin/env Rscript
+# Question:
+# In table 5 it is shown that the application
+# is thought of as more useful and they
+# give statistics. Can this be reproduced?
+
+before_mean <- 4.78
+before_sd <- 0.67
+
+after_mean <- 5.67
+after_sd <- 0.50
+
+expected_p_value <- 0.011
+n_people <- 9
+n_p_values <- 100
+n_experiments <- 10
+
+simulate_p_values <- function(
+    before_mean = 4.78,
+    before_sd = 0.67,
+    after_mean = 5.67,
+    after_sd = 0.50,
+    n_people = 9,
+    n_p_values = 100
+) {
+  testthat::expect_equal(before_mean, 4.78)
+  testthat::expect_equal(before_sd, 0.67)
+  testthat::expect_equal(after_mean, 5.67)
+  testthat::expect_equal(after_sd, 0.5)
+  testthat::expect_equal(n_people, 9)
+  testthat::expect_equal(n_p_values, 100)
+
+  p_values <- rep(NA, n_p_values)
+  for (i in seq_len(n_p_values)) {
+    dist_before <- rnorm(n = n_people, mean = before_mean, sd = before_sd)
+    dist_after <- rnorm(n = n_people, mean = after_mean, sd = after_sd)
+    stat_results <- t.test(x = dist_before, y = dist_after)
+    p_values[i] <- stat_results$p.value
+  }
+  testthat::expect_equal(0, sum(is.na(p_values)))
+  p_values
+}
+
+
+all_data <- list()
+for (i in seq_len(n_experiments)) {
+  t <- tibble::tibble(
+    p = simulate_p_values(n_p_values = n_p_values)
+  )
+  t$i <- i
+  all_data[[i]] <- t
+}
+t <- dplyr::bind_rows(all_data)
+
+
+ggplot2::ggplot(t, ggplot2::aes(x = p)) +
+  ggplot2::geom_histogram() + 
+  ggplot2::facet_wrap(ggplot2::vars(i), ncol = 2) +
+  ggplot2::scale_x_continuous(limits = c(0.0, 0.06)) +
+  ggplot2::geom_vline(xintercept = expected_p_value) +
+  ggplot2::geom_vline(xintercept = 0.05, lty = "dashed") +
+  ggplot2::geom_vline(xintercept = mean(t$p), color = "red") + 
+  ggplot2::labs(
+    title = "Distribution of simulated p-values",
+    caption = paste0(
+        "Black vertical line: reported in paper. \n",
+        "Black dashed vertical line: 0.05%. \n",
+        "Red vertical line: mean p value from ", n_p_values, " simulations. "
+      )
+    )
+
+ggplot2::ggsave("20241011_tabel_5_sim.png", width = 4, height = 7)

20241016.md

@@ -0,0 +1,29 @@
+# 2024-10-16: Lockman & Schirmer, 2020
+
+- Discussion leader: Stephan
+- Date: 2024-10-16
+- Paper: Dimitriadou, Eleni, and Andreas Lanitis. "Evaluating the impact of an automated body language assessment system." Education and Information Technologies (2024): 1-31.
+
+First sentence already is deceptive: 'Body language is an important aspect in
+educational settings, directly influencing
+educators’ effectiveness in message delivery, classroom management, and student
+interactions (Benzer, 2012).'. This references a paper called
+'Teachers’ Opinions about the Use of Body Language', where
+that ask 100 teachers
+if they think body language is important.
+'When asked, teachers think body language is an important aspect in
+educational settings, with possible effects on influencing
+educators’ effectiveness in message delivery, classroom management, and student
+interactions'.
+
+Sample size of 9
+
+Table 5: 'Do you think the application
+is useful for improving the
+lecture quality?' before and after using
+the tool.
+
+4.78 (0.67) 0.011*
+Do you think the application
+is useful for improving the
+lecture quality?