07-s01-lab07e.Rmd

## Solutions to Questions

```{r lab7-solutions-setup, echo = FALSE, message=FALSE, warning=FALSE}
library(broom)
ratings <- read_csv("data/07-s01/inclass/GuiltJudgements.csv")
```

Below you will find the solutions to the questions for the Activities for this chapter. Only look at them after giving the questions a good try and speaking to the tutor about any issues.

### InClass Activities

#### InClass Task 1

```{r ch7-task1, eval = FALSE}
library(broom)
library(tidyverse)
ratings <- read_csv("GuiltJudgements.csv")
```

[Return to Task](#Ch7InClassQueT1)

#### InClass Task 2

```{r ch7-task2, results='asis'}
lates <- ratings %>% 
  filter(Evidence == "Late") %>%
  select(Participant, Evidence, `12`, `13`) %>%
  rename(Twelve = `12`, Thirteen = `13`) %>%
  mutate(diff = Twelve - Thirteen)
```

[Return to Task](#Ch7InClassQueT2)

#### InClass Task 3

```{r ch7-task3, results='asis', fig.cap = "Potential Solution to Task 3"}
lates %>% 
  ggplot(aes(diff)) +
  geom_histogram(binwidth = 1) +
  labs(x = "Difference in GuiltRating", y = NULL) +
  theme_bw()
```

[Return to Task](#Ch7InClassQueT3)

#### InClass Task 4

The Task only asks for the boxplot. We have added some additional functions to tidy up the figure a bit that you might want to play with.

```{r ch7-task4, results='asis', fig.cap = "Potential Solution to Task 4"}
lates %>% 
  ggplot(aes(y = diff)) + 
  geom_boxplot() +
  theme_bw()
```

[Return to Task](#Ch7InClassQueT4)

#### InClass Task 5

```{r ch7-task5, results='asis'}
descriptives <- lates %>% 
  select(-diff) %>%
  gather(Timepoint, GuiltRating, Twelve, Thirteen) %>%
  group_by(Timepoint) %>%
  summarise(n = n(),
            mean = mean(GuiltRating),
            sd = sd(GuiltRating),
            se = sd/sqrt(n),
            LowerCI = mean - 1.96*se,
            UpperCI = mean + 1.96*se)
```

```{r hidden-1, include = FALSE}
desc2 <- descriptives %>%
  mutate_at(vars(mean, sd, se, LowerCI, UpperCI),
            webex::round2, digits = 2)
.d12 <- desc2 %>% filter(Timepoint == "Twelve")
.d13 <- desc2 %>% filter(Timepoint == "Thirteen")
```


[Return to Task](#Ch7InClassQueT5)

#### InClass Task 6

* A basic barplot with 95% Confidence Intervals. 
* We have embellished the figure a little but you can mess around with the code to see what each bit does.

```{r ch7-task6-1, results='asis', fig.cap = "Possible Solution to Task 7"}
ggplot(descriptives, aes(x = Timepoint, y = mean, fill = Timepoint)) + 
  geom_col(colour = "black") +
  scale_fill_manual(values=c("#E2A458", "#61589C")) +
  scale_x_discrete(limits = c("Twelve","Thirteen")) +
  labs(x = "Timepoint of Evidence", y = "GuiltRating") +
  guides(fill="none") +
  geom_errorbar(aes(ymin = LowerCI, ymax = UpperCI),
                position = "dodge", width = .15) +
  scale_y_continuous(breaks = c(1:9), limits = c(0,9)) +
  coord_cartesian(ylim = c(1,9), xlim = c(0.5,2.5), expand = FALSE) +
  theme_classic()
```

* An alternative way to display just the means and errorbars would be to use the pointrange approach. This image shows again the 95% CI

```{r ch7-task6-2, results='asis', fig.cap = "Alternative Solution to Task 7"}
ggplot(descriptives, aes(x = Timepoint, y = mean, fill = Timepoint)) + 
  geom_pointrange(aes(ymin = LowerCI, ymax = UpperCI))+
  scale_x_discrete(limits = c("Twelve","Thirteen")) +
  labs(x = "Timepoint of Evidence", y = "GuiltRating") +
  guides(fill="none")+
  scale_y_continuous(breaks = c(1:9), limits = c(0,9)) +
  coord_cartesian(ylim = c(1,9), xlim = c(0.5,2.5), expand = FALSE) +
  theme_bw()
```

[Return to Task](#Ch7InClassQueT6)

#### InClass Task 7

* Remember to set `paired = TRUE` to run the within-subjects t-test

```{r ch7-task7, results='asis'}
results <- t.test(lates %>% pull(Twelve),
                  lates %>% pull(Thirteen), paired = TRUE) %>%
  tidy()
```

[Return to Task](#Ch7InClassQueT7)

#### InClass Task 8

A potential write-up for this study would be as follows:

**A paired-samples t-test was run to compare the change in guilt ratings before (M = `r .d12$mean`, SD = `r .d12$sd`) and after (M = `r .d13$mean`, SD = `r .d13$sd`) the crucial evidence was heard. A significant difference was found (t(`r webex::round2(results$parameter,2)`) = `r webex::round2(results$statistic, 2)`, p `r if_else(webex::round2(results$p.value, 3) < .001, "< .001", paste0("= ", webex::round2(results$p.value, 3)))`) with Timepoint 13 having an average rating `r webex::round2(results$estimate, 2) %>% abs()` units lower than Timepoint 12. This tells us that the critical evidence did have an influence on the rating of guilt by jury members.**

When rounding off p-values that are less than .001, rounding will give you a value of 0 which is a bit silly. Values less than .001 would normally be written as p < .001. To create a reader-friendly p-value, then you could try something like the following in your code:

```{r example-1, eval=FALSE}
ifelse(results$p.value < .001, 
       "p < .001", 
       paste0("p = ", round(results$p.value,3))) 
```

[Return to Task](#Ch7InClassQueT8)

<span style="font-size: 22px; font-weight: bold; color: var(--purple);">Chapter Complete!</span>