analyze_classroom_pgg2020.Rmd

---
title: "Results classroom PGG"
author: "Jantsje Mol"
date: "23/9/2020"
output: html_document
---

```{r setup, include=FALSE}
# !diagnostics off
here <- here::here # because lubridate also has a here function

library(tidyverse, stringr)
library(numbers)
library(dplyr)
library(anytime) # for converting Epoch times from oTree Page Time 

select <- dplyr::select
filter <- dplyr::filter

timespent1 <- read.csv(here("_DATA", "PageTimes-2020-11-24.csv"), sep = ',', header=TRUE)
timespent2 <- read.csv(here("_DATA", "PageTimes-2020-11-26.csv"), sep = ',', header=TRUE)
timespent <- rbind(timespent1, timespent2) %>% 
  rename(pp_code = participant_code) %>% 
  mutate(timestamp = anytime(epoch_time)) %>% 
  add_count(pp_code, sort=TRUE) %>% 
  filter(app_name == "classroom_pgg") %>% 
  filter(page_index == 1 | page_index == 70) %>% 
  select(c(pp_code, page_index, timestamp))

rm(timespent1, timespent2)

timespent <- pivot_wider(timespent, names_from = page_index, names_prefix = "page", values_from = timestamp) %>% 
  filter(!is.na(page70))
timespent <- timespent %>% 
  mutate(seconds_total = difftime(page70, page1, units = "secs"))%>% 
  mutate(pp_minutes= as.integer(as.numeric(seconds_total)/60),
         pp_seconds= mod(as.numeric(seconds_total),60)) %>% 
  select(-starts_with("page"))

df1 <- read.csv(here("_DATA", "all_apps_wide_2020-11-24.csv"), sep = ',', header=TRUE) %>% 
  filter(participant._current_app_name == "classroom_pgg")

df2 <- read.csv(here("_DATA", "all_apps_wide_2020-11-26.csv"), sep = ',', header=TRUE) %>% 
  filter(participant._current_app_name == "classroom_pgg")

df <- rbind(df1, df2) %>% 
  rename(pp_code = participant.code,
         pp_label = participant.label,
         pp_payoff = participant.payoff,
         pp_start_time = participant.time_started,
         pp_id = participant.id_in_session,
         session_label = session.label,
         grup_id = classroom_pgg.1.group.id_in_subsession) # typo is intentional to be able to deselect all other group_ variables
rm(df1, df2)

df <- merge(df, timespent, by="pp_code", all.x = TRUE) %>% 
  select(-pp_code)
rm(timespent)

df <- df %>% 
  rename(session_num_rounds = session.config.num_rounds,
         session_chat = session.config.cheap_talk,
         session_threshold = session.config.threshold,
         session_punishment = session.config.punishment) %>% 
  select(-contains("config")) %>%
  select(-contains("floodgame")) %>% 
  select(-contains("punishment_endowment")) %>%
  select(-contains("subsession")) %>%
  select(-contains("id_in_group")) %>%
  select(-contains("_share")) %>%
  select(-contains("average_contribution")) %>%
  select(-contains("pd_payoff")) %>%
  select(-c(participant._is_bot, participant._max_page_index,
            participant._current_page_name, participant._current_app_name, participant.visited, 
            participant.mturk_worker_id, participant.mturk_assignment_id, session.mturk_HITId, 
            session.comment, session.mturk_HITGroupId, session.is_demo, participant._index_in_pages,
            participant.payoff_plus_participation_fee)) 

thenames <- names(df)
function.clean_varnames <-  function(text){
  if(grepl("player", text, perl=TRUE) | grepl("group", text, perl=TRUE)){
    if(grepl("classroom_pgg.", text, perl=TRUE)){text <- sub(pattern = "classroom_pgg.", replacement = "", x = text)}
    if(grepl(".player", text, perl=TRUE)){text <- sub(pattern = ".player", replacement = "", x = text)} 
    if(grepl(".group", text, perl=TRUE)){text <- sub(pattern = ".group", replacement = "", x = text)} 
    text <- strsplit(text, '[.]')
    text <- text[[1]][c(2,1)]
    text <- paste(text, collapse='_')
  }
  else {text <- text}} 
thenames <- lapply(thenames, function.clean_varnames)
names(df) <- thenames 
rm(thenames, function.clean_varnames)


df <- df %>% 
  filter(session_label != "test-jantsje") %>% 
  filter(pp_label != "") %>% # used to force to next page
  select(starts_with("pp_"), starts_with("session"), grup_id, ends_with("_1"), ends_with("_2"), ends_with("_3"), ends_with("_4"),
         ends_with("_5"), ends_with("_6"), ends_with("_7"), ends_with("_8"), ends_with("_9"), ends_with("_10")) %>% 
  gather(variable,value,-c("pp_label", "pp_start_time", "pp_payoff", "pp_id", "session.code", "session_label",
                           "pp_minutes", "pp_seconds", 
                           "session_chat", "session_threshold", "session_punishment", "session_num_rounds",
                           "grup_id")) %>%
  mutate(round = parse_number(variable)) %>%
  mutate(variable =  gsub("\\d","",x = variable)) %>% 
  spread(variable,value) %>% 
  rename(contribution = contribution_,
         payoff = payoff_,
         group_contribution = total_contribution_,
         reached_threshold = reached_threshold_,
         punishment_received = punishment_received_,
         punishment_sent = punishment_sent_,
         session_code = session.code,
         group_id = grup_id) %>% 
  mutate(contribution = case_when(round > session_num_rounds ~ NA_integer_,
                                  TRUE ~ contribution),
         payoff = case_when(round > session_num_rounds ~ NA_integer_,
                                  TRUE ~ payoff),
         punishment_received = case_when(round > session_num_rounds ~ NA_integer_,
                                  TRUE ~ punishment_received),
         punishment_sent = case_when(round > session_num_rounds ~ NA_integer_,
                                  TRUE ~ punishment_sent),
         reached_threshold = case_when(round > session_num_rounds ~ NA_integer_,
                                     TRUE ~ reached_threshold),
         group_contribution = case_when(round > session_num_rounds ~ NA_integer_,
                                     TRUE ~ group_contribution),
         punishment_received = case_when(session_punishment == 0 ~ NA_integer_,
                                         TRUE ~ punishment_received),
         punishment_sent = case_when(session_punishment == 0 ~ NA_integer_,
                                         TRUE ~ punishment_sent),
         reached_threshold = case_when(session_threshold == 0 ~ NA_integer_,
                                     TRUE ~ reached_threshold)) 

parameters.session1 <- df %>% 
  filter(session_label == "session 1") %>% 
  filter(pp_id == max(pp_id), round == 1) %>% 
  select(starts_with("session"), pp_id, group_id)
```

## Session 1: one shot PGG
*Monday November 2nd, 2020*

In this game, `r parameters.session1$pp_id` students participated.

* Number of rounds: `r parameters.session1$session_num_rounds`
* Group size: 3
* Multiplication factor: 1.6
* Punishment: `r parameters.session1$session_punishment`
* Threshold: `r parameters.session1$session_threshold`
* Chat: `r parameters.session1$session_chat`

```{r histograms contributions, echo = FALSE, warning = FALSE, message=FALSE}
df.session1 <- df %>% 
  filter(round == 1,
         session_label == "session 1") 

fig.contribution <- ggplot(df.session1, aes(x=contribution), subset = round == 1) + geom_histogram(binwidth = 5, color = "white") +
  labs(x="Individual contribution", y = "Frequency") 

fig.group.contribution <- ggplot(df.session1, aes(x=group_contribution), subset = round == 1) + geom_histogram(binwidth = 10, color = "white") +
  labs(x="Group contribution", y = "Frequency") + scale_x_continuous(limits = c(0,300))

library(gridExtra)
grid.arrange(fig.contribution, fig.group.contribution,
             nrow = 1)

ggplot(df.session1, aes(x=payoff)) + geom_histogram(binwidth = 5, color = "white") +
  labs(x="Individual payoff in points", y = "Frequency") + scale_x_continuous(limits = c(40,220)) +
  geom_vline(xintercept=160/3, linetype = "dashed") + annotate(geom = "label", x = 160/3, y = 13, label = "minimum") +
  geom_vline(xintercept=206, linetype = "dashed") + annotate(geom = "label", x = 205, y = 13, label = "maximum") +
  geom_vline(xintercept=(300*1.6)/3, linetype = "dashed") + annotate(geom = "label", x = (300*1.6)/3, y = 13, label = "full contribution") 

rm(list = ls(pattern = "fig"))
rm(parameters.session1)

```

```{r, include = FALSE}
parameters.session2 <- df %>% 
  filter(session_label == "session 2") %>% 
  filter(pp_id == max(pp_id), round == 1) %>% 
  select(starts_with("session"), pp_id, group_id)

```
## Session 2: ten rounds PGG
*Wednesday November 11th, 2020*

In this game, `r parameters.session2$pp_id` students participated.

* Number of rounds: `r parameters.session2$session_num_rounds`
* Group size: 3
* Multiplication factor: 1.6
* Punishment: `r parameters.session2$session_punishment`
* Threshold: `r parameters.session2$session_threshold`
* Chat: `r parameters.session2$session_chat`

```{r, include = FALSE}
parameters.session3 <- df %>% 
  filter(session_label == "session 3") %>% 
  filter(pp_id == max(pp_id), round == 1) %>% 
  select(starts_with("session"), pp_id, group_id)

```
## Session 3: ten rounds PGG + threshold
*Thursday November 19th, 2020*

In this game, `r parameters.session3$pp_id` students participated.

* Number of rounds: `r parameters.session3$session_num_rounds`
* Group size: 3
* Multiplication factor: 1.6
* Punishment: `r parameters.session3$session_punishment`
* Threshold: `r parameters.session3$session_threshold`
* Chat: `r parameters.session3$session_chat`

```{r, include = FALSE}
parameters.session4 <- df %>% 
  filter(session_label == "session 4") %>% 
  filter(pp_id == max(pp_id), round == 1) %>% 
  select(starts_with("session"), pp_id, group_id)

```

```{r clean chat data}
library(stringr)

chats <- read.csv(here("_DATA", "ChatMessages-2020-11-26.csv"), sep = ',', header=TRUE) %>% 
  select(c("participant__code", "nickname", "channel", "body", "timestamp")) %>% 
  mutate(timestamp = anytime(timestamp)) %>% 
  rename(participant = participant__code) %>% 
  mutate(channel = as.character(channel)) %>% 
  mutate(channel = as.integer(str_remove(channel, "3-classroom_pgg-"))) %>% 
  mutate(group = channel%%70-30) %>% 
  mutate(group = case_when(group < 0 ~ group + 70, TRUE ~ group),
         round = as.integer(channel/70))

chat_sum <- chats %>% 
  group_by(group, round) %>% 
  tally()

ggplot(chat_sum, aes(x=round)) + 
  geom_line(data = subset(chat_sum, group == 1), aes(y=n), color = "blue", alpha = 0.5) +
  geom_jitter(aes(y=n), width = 0.2) +
  scale_x_continuous(breaks = c(1:10)) + 
  labs(y="Number of chat messages", x= "Round") 

ggplot(chat_sum) + 
  geom_boxplot(aes(y = n, group = factor(round))) +
  scale_x_continuous(breaks = c(1:10), labels = c(1:10)) +
  labs(y="Number of chat messages", x= "Round") 


```


## Session 4: ten rounds PGG
*Thursday November 26th, 2020*

In this game, `r parameters.session4$pp_id` students participated.

* Number of rounds: `r parameters.session4$session_num_rounds`
* Group size: 3
* Multiplication factor: 1.6
* Punishment: `r parameters.session4$session_punishment`
* Threshold: `r parameters.session4$session_threshold`
* Chat: `r parameters.session4$session_chat`




## Latest leaderboard
```{r, echo = FALSE, warning= FALSE, message=FALSE, results = 'asis'}
 payoffs <- df %>%
  filter(round == 10) %>% 
  select(pp_label, pp_payoff, session_label) %>% 
  pivot_longer(cols = pp_payoff) %>% 
   filter(pp_label != "mke240") %>% # teachers cannot win
   filter(pp_label != "aaa111") %>%
   filter(pp_label != "jml740") %>% 
  group_by(pp_label) %>% 
  mutate(pp_payoff_total = sum(value))

df.leaderboard <- payoffs %>% 
  ungroup() %>% 
  filter(session_label == "session 1") %>% 
   arrange(-pp_payoff_total) %>%
   filter(row_number() < 4) %>%
   mutate(ranking = row_number()) %>%
   select(ranking, pp_label, pp_payoff_total) # this also adds the order

 library(kableExtra)
 knitr::kable(df.leaderboard) %>%
   kableExtra::kable_styling(full_width = F,
                             position = "left")
```



## Average individual contributions over rounds

```{r, echo = FALSE, warning= FALSE, message=FALSE}
df <- df %>% 
  mutate(session_label_inf = case_when(session_label == "session 1" ~ "Basic one round",
                                       session_label == "session 2" ~ "Basic ten rounds",
                                       session_label == "session 3" ~ "Threshold",
                                       session_label == "session 4" ~ "Threshold + chat",
                                       session_label == "session 5" ~ "Threshold + chat + punishment"))

df.gather <- df %>% 
  group_by(round, session_label_inf) %>% 
  summarise(mean.contribution = mean(contribution, na.rm=TRUE),
            sd.contribution = sd(contribution, na.rm=TRUE),
            n.contribution = n()) %>% 
  mutate(se.contribution = sd.contribution / sqrt(n.contribution),
         lower.ci.contribution = mean.contribution - qt(1 - (0.05 / 2), n.contribution - 1) * se.contribution,
         upper.ci.contribution = mean.contribution + qt(1 - (0.05 / 2), n.contribution - 1) * se.contribution)


ggplot(data=subset(df.gather, session_label_inf == "Basic one round"), aes(x=round, y=mean.contribution, color = session_label_inf)) + geom_point() + geom_line() +
  geom_errorbar(aes(ymin=lower.ci.contribution, ymax=upper.ci.contribution), alpha=0.9, width=.1) +
  scale_x_continuous(breaks = c(1:10)) + labs(y="Average individual contribution", x= "Round", color = "Session") +
  scale_y_continuous(limit = c(0,100)) + scale_color_brewer(palette="Dark2")

ggplot(data = subset(df.gather, session_label_inf == "Basic one round" | session_label_inf ==  "Basic ten rounds"), aes(x=round, y=mean.contribution, color = session_label_inf)) + geom_point() + geom_line() +
  geom_errorbar(aes(ymin=lower.ci.contribution, ymax=upper.ci.contribution), alpha=0.9, width=.1) +
  scale_x_continuous(breaks = c(1:10)) + labs(y="Average individual contribution", x= "Round", color = "Session") +
  scale_y_continuous(limit = c(0,100)) + scale_color_brewer(palette="Dark2")

ggplot(data = subset(df.gather, 
                     session_label_inf == "Basic one round" | 
                       session_label_inf ==  "Basic ten rounds" | 
                       session_label_inf == "Threshold"), aes(x=round, y=mean.contribution, color = session_label_inf)) + geom_point() + geom_line() +
  geom_errorbar(aes(ymin=lower.ci.contribution, ymax=upper.ci.contribution), alpha=0.9, width=.1) +
  scale_x_continuous(breaks = c(1:10)) + labs(y="Average individual contribution", x= "Round", color = "Session") +
  scale_y_continuous(limit = c(0,100)) + scale_color_brewer(palette="Dark2")

ggplot(data = subset(df.gather, 
                     session_label_inf == "Basic one round" | 
                       session_label_inf ==  "Basic ten rounds" | 
                       session_label_inf == "Threshold" | 
                       session_label_inf == "Threshold + chat"), aes(x=round, y=mean.contribution, color = session_label_inf)) + geom_point() + geom_line() +
  geom_errorbar(aes(ymin=lower.ci.contribution, ymax=upper.ci.contribution), alpha=0.9, width=.1) +
  scale_x_continuous(breaks = c(1:10)) + labs(y="Average individual contribution", x= "Round", color = "Session") +
  scale_y_continuous(limit = c(0,100)) + scale_color_brewer(palette="Dark2")

ggplot(df.gather, aes(x=round, y=mean.contribution, color = session_label_inf)) + geom_point() + geom_line() +
  geom_errorbar(aes(ymin=lower.ci.contribution, ymax=upper.ci.contribution), alpha=0.9, width=.1) +
  scale_x_continuous(breaks = c(1:10)) + labs(y="Average individual contribution", x= "Round", color = "Session") +
  scale_y_continuous(limit = c(0,100)) + scale_color_brewer(palette="Dark2")

```