Early Course Evaluation - STA 418/518

Preparing R

library(tidyverse)
library(googlesheets4)
library(RColorBrewer)
library(tidytext)
library(ggwordcloud)

# You wont be able to run this as you do not have access on my GDrive
# However, the string here is from the Google Sheet's url - between the "/d/ and "/edit..."
sta518_evals <- read_sheet("1MkwRbzcshJo1qPzooP2Ww829KzUN2jj4EH2eKPCWRfA",
# Like the tidyverse "read_*" functions, you can specify missing values.
                           na = "I have not attempted")
# All questions were required so I am treating students that have not attempted something as missing.

likert_scale <- c("Strongly disagree", "Slightly disagree",
                  "Slightly agree", "Strongly agree")

likert_pallet <- RColorBrewer::brewer.pal(length(likert_scale), "BrBG")

likert_info <- tibble(
  scale = factor(likert_scale, levels = likert_scale),
  color = likert_pallet
)

Course Response Breakdown

sta518_evals %>% 
  select(course = starts_with("In which c")) %>% 
  ggplot(aes(y = course)) +
  geom_bar(fill = "#0065a4") +
  theme_bw()

Course Design

# Rel freq table by likert response
course_design_summary <- sta518_evals %>% 
  select(starts_with("Course")) %>% 
  mutate(across(everything(), ~factor(., level = likert_scale))) %>% 
  pivot_longer(everything(), names_to = "item", values_to = "response",
               names_pattern = "Course design \\[(.*)\\]") %>% 
  group_by(item, response) %>% 
  summarise(freq = n()) %>% 
  mutate(perc = freq / sum(freq) * 100)

# Add color column
course_design <- course_design_summary %>% 
  left_join(likert_info, by = c("response" = "scale"))

# Select only agrees
course_design_agree <- course_design %>% 
  filter(str_detect(response, " agree"))

# Select only disagrees
course_design_disagree <- course_design %>% 
  filter(!str_detect(response, " agree"))

# PLOT!!!!
ggplot() +
  # Notice that I am plotting two different datasets, the "agree" dataset:
  geom_bar(data = course_design_agree, 
           aes(y = str_wrap(item, width = 40), x = perc, fill = color),
           position = "stack", stat = "identity") +
  # And the "disagree" dataset:
  geom_bar(data = course_design_disagree,
           aes(y = str_wrap(item, width = 40), x = -perc, fill = color),
           position = "stack", stat = "identity") +
  labs(y = "",
       x = "Percent") +
  scale_x_continuous(limits = c(-100, 100)) +
  scale_fill_identity("", 
                    labels = likert_scale, 
                    guide = "legend",
                    breaks = likert_pallet) +
  theme_bw() +
  theme(legend.position="bottom",
        legend.box="horizontal",
        legend.text = element_text(size = 8))

Instructor

instructor_summary <- sta518_evals %>% 
  select(starts_with("Instructor")) %>% 
  mutate(across(everything(), ~factor(., level = likert_scale))) %>% 
  pivot_longer(everything(), names_to = "item", values_to = "response",
               names_pattern = "Instructor \\[(.*)\\]") %>% 
  group_by(item, response) %>% 
  summarise(freq = n()) %>% 
  mutate(perc = round(freq / sum(freq) * 100, 1))

instructor <- instructor_summary %>% 
  select(-freq) %>% 
  left_join(likert_info, by = c("response" = "scale"))

instructor_agree <- instructor %>% 
  filter(str_detect(response, " agree"))

instructor_disagree <- instructor %>% 
  filter(!str_detect(response, " agree"))

ggplot() +
  geom_bar(data = instructor_agree,
           aes(y = str_wrap(item, width = 40), x = perc, fill = color),
           position = "stack", stat = "identity") +
  geom_bar(data = instructor_disagree,
           aes(y = str_wrap(item, width = 40), x = -perc, fill = color),
           position = "stack", stat = "identity") +
  labs(y = "",
       x = "Percent") +
  scale_x_continuous(limits = c(-100, 100)) +
  scale_fill_identity("", labels = likert_scale, 
                      breaks = likert_pallet, guide = "legend",
                      drop = FALSE) +
  theme_bw() +
  theme(legend.position="bottom",
        legend.box="horizontal",
        legend.text = element_text(size = 8))

Free response

What has been the best aspect of this course so far?

best <- sta518_evals %>% 
  select(best = starts_with("What as")) %>% 
  mutate(individual = row_number()) %>% 
  unnest_tokens(bigram, best, token = "ngrams", n = 2)

best_sing <- sta518_evals %>% 
  select(starts_with("What as")) %>% 
  rename(best = starts_with("What as")) %>% 
  mutate(individual = row_number()) %>% 
  unnest_tokens(bigram, best) %>% 
  anti_join(get_stopwords(), by = c("bigram" = "word"))

best_sep <- best %>%
  separate(bigram, c("word1","word2"), sep = " ")

best_filtered <-  best_sep %>% 
  filter(!word1 %in% stop_words$word) %>% 
  filter(!word2 %in% stop_words$word)

best_cleaned <- best_filtered %>% 
  unite(bigram, word1, word2, sep = " ")

best_cleaned <- bind_rows(best_sing, best_cleaned)

best_cleaned %>%
  count(bigram, sort = TRUE) %>%
  filter(n > 1) %>% 
  ggplot(aes(label = bigram, size = n)) + 
  geom_text_wordcloud() +
  scale_size_area(max_size = 10) +
  theme_bw()

If there was one thing about this course that you would change?

change <- sta518_evals %>% 
  select(change = starts_with("If there")) %>% 
  mutate(individual = row_number()) %>% 
  unnest_tokens(bigram, change, token = "ngrams", n = 2)

change_sing <- sta518_evals %>%
  select(starts_with("If there")) %>% 
  rename(change = starts_with("If there")) %>% 
  mutate(individual = row_number()) %>% 
  unnest_tokens(bigram, change) %>% 
  anti_join(get_stopwords(), by = c("bigram" = "word"))

change_sep <- change %>%
  separate(bigram, c("word1","word2"), sep = " ")

change_filtered <-  change_sep %>% 
  filter(!word1 %in% stop_words$word) %>% 
  filter(!word2 %in% stop_words$word)

change_cleaned <- change_filtered %>% 
  unite(bigram, word1, word2, sep = " ")

change_cleaned <- bind_rows(change_sing, change_cleaned)

change_cleaned %>%
  count(bigram, sort = TRUE) %>%
  filter(n > 1) %>% 
  ggplot(aes(label = bigram, size = n)) + 
  geom_text_wordcloud() +
  scale_size_area(max_size = 10) +
  theme_bw()

What is one thing that you can do differently to improve your learning?

improve <- sta518_evals %>% 
  select(differently = starts_with("What is")) %>% 
  mutate(individual = row_number()) %>% 
  unnest_tokens(bigram, differently, token = "ngrams", n = 2)

improve_sing <- sta518_evals %>% 
  select(differently = starts_with("What is")) %>% 
  mutate(individual = row_number()) %>% 
  unnest_tokens(bigram, differently) %>% 
  anti_join(get_stopwords(), by = c("bigram" = "word"))

improve_sep <- improve %>%
  separate(bigram, c("word1","word2"), sep = " ")

improve_filtered <-  improve_sep %>% 
  filter(!word1 %in% stop_words$word) %>% 
  filter(!word2 %in% stop_words$word)

improve_cleaned <- improve_filtered %>% 
  unite(bigram, word1, word2, sep = " ")

improve_cleaned <- bind_rows(improve_sing, improve_cleaned)

improve_cleaned %>%
  count(bigram, sort = TRUE) %>%
  filter(n > 1) %>% 
  ggplot(aes(label = bigram, size = n)) + 
  geom_text_wordcloud() +
  scale_size_area(max_size = 10) +
  theme_bw()