evaluation_of_demonstrator.Rmd

---
output:
  html_document:
    theme: spacelab
---

# Evaluation of the demonstrator and the observational US

This analysis investigates the evaluation of the demonstrators by the observers.

```{r, message=FALSE}
library('tidyverse')
library('ggpubr')
library('psych')
library('ini')
```

We are working with the following files
```{r}
config <- read.ini("../config.ini")
tab_file <- file.path(config$DEFAULT$QUESTIONNAIRE_DIR, 'table.tsv')
```

The analysis is based on the observers from the friend group:

```{r}
df_ca <- read_tsv(tab_file, col_types=cols()) %>%
    select(
        c("subject_id", "label", "role", "group", "demonstrator_id", starts_with("ED"), CONT_shock_pleasant)
    ) %>%
    filter(role == "OBS") %>%
    filter(group == "friend")
```

The ratings can be plotted:

```{r, warning=FALSE}
ed_plot <- df_ca %>%
    select(starts_with("ED")) %>%
    pivot_longer(cols = starts_with("ED")) %>%  #creates name, value
    mutate(
        # change the variables names & convert to factor
        name = as_factor(str_replace(name, "ED_", ""))
    ) %>%
    mutate(
        # change order and then names to control how they appear in the figure
        name = fct_recode(
            fct_relevel(name, c("discomfort", "expressive", "natural", "empathy", "identify")),
            expressiveness = "expressive",
            naturalness = "natural",
            identifying = "identify"
        )
    ) %>%
    ggdotplot(x = "name", y = "value", facet.by = "name",
              dotsize = 0.4, alpha = 0.35, nrow = 1, ncolumn = 5, font.label = 1) %>%
    ggadd("boxplot", alpha = 0) %>%            # this way, boxplot's on top
    ggpar(yticks.by = 1, font.ytickslab = 8, palette = c("#D55E00", "#0072B2")) %>%
    + rremove("x.text") + rremove("x.ticks") + rremove("xlab") + theme(strip.text.x = element_text(size = 8))


if(!dir.exists("figures")) {
    dir.create("figures")
}

target_ppi = 500

ggexport(
    ed_plot,
    filename = "figures/evaluation_of_demonstrator_dotplot.tiff",
    width = 5.51 * target_ppi,
    height = 3.86 * target_ppi,
    res = target_ppi,
    pointsize = 8
)
ed_plot
```

Summary statistics for the ratings, calculated using `describeBy` from the `psych` package:

```{r, warning=FALSE}
dsg <- df_ca %>%
    select(where(is.numeric)) %>%
    describeBy(IQR=TRUE)
dsg
```