---
title: "R Workshop 2: Visualizing and Describing Data"
subtitle: "Student workbook"
format:
  html:
    toc: true
    toc-depth: 2
execute:
  echo: true
  warning: false
  message: false
---

## What you will be able to do

By the end of today, you will be able to:

- use the native R pipe `|>` to organize a series of computations;
- identify categorical and numeric variables;
- choose a graph that matches the variables you want to study;
- create bar charts, histograms, scatterplots, and boxplots with `ggplot2`;
- calculate counts, proportions, and numerical descriptive statistics;
- calculate grouped summaries and a descriptive correlation; and
- combine graphs, statistics, and prose in a reproducible Quarto document.


Demonstration chunks in this workbook run when the document renders. Practice chunks contain `#| eval: false`, so an unfinished exercise will not prevent the document from rendering. You can run practice code interactively with the Run button while you work.

## 1. Reopen the project and import the data

Open `R-Workshop-1.Rproj`, then open this file from RStudio’s **Files** pane. Opening the project gives R a stable starting point for relative paths.

Load the tidyverse packages:

```{r}
library(tidyverse)
```

Import the CSV with a project-relative path:

```{r}
dat <- read_csv("data/workshop1_survey.csv")
```

Confirm that the data frame exists and inspect its structure:

```{r}
glimpse(dat)
nrow(dat)
ncol(dat)
```

One row represents one synthetic undergraduate participant. One column represents one variable.

### Counts and the pipe

The native R pipe `|>` passes the result on its left into the operation on the next line.

```{r}
dat |>
  count(relationship_status)
```

Read this from top to bottom: “Take `dat`, then count the observations in each relationship-status category.” The `NA` row represents participants whose relationship status is missing.

To insert `|>` in RStudio, use **Windows: Ctrl+Shift+M** or **Mac: Shift+Command+M**.

### Your turn: copy and change one variable

Copy the preceding pipeline into this chunk and change `relationship_status` to `gender`.

```{r}
#| eval: false

```

### Your turn: fill in the blank

Count each value of `leadership_interest`. This is a numeric scale, so notice how many distinct values appear.

```{r}
#| eval: false
dat |>
  count(_____)
```

## 2. Variables determine the graph

A **categorical variable** places observations into categories. Examples include `gender` and `relationship_status`.

A **numeric variable** records quantities for which arithmetic is meaningful. Examples include `age`, `height_cm`, `extraversion`, and `social_confidence`.

| Variables in the question | Useful first graph |
|---|---|
| One categorical variable | Bar chart |
| One numeric variable | Histogram |
| Two numeric variables | Scatterplot |
| One categorical and one numeric variable | Boxplot |

This table is a starting point, not a rule for every analysis.

### Your turn: identify variable types

For each variable, write **categorical** or **numeric**.

- `relationship_status`: **[Write here]**
- `openness`: **[Write here]**
- `gender`: **[Write here]**
- `number_close_friends`: **[Write here]**

Which graph would you begin with for each question?

1. How are participants distributed across relationship-status categories? **[Write here]**
2. What does the distribution of openness scores look like? **[Write here]**
3. How are extraversion and social confidence related? **[Write here]**

## 3. Visualizing categorical variables

### A bar chart

`ggplot()` begins the graph. Inside `aes()`, we state which variables play which visual roles. A **geom** determines what kind of marks the graph uses.

```{r}
ggplot(data = dat, aes(x = relationship_status)) +
  geom_bar() +
  labs(
    title = "Relationship status in the synthetic sample",
    x = "Relationship status",
    y = "Number of participants"
  ) +
  theme_minimal()
```

Layers in a `ggplot` are joined with `+`. This is different from the pipe used to pass data through a sequence of operations.

Before calculating anything, describe the graph:

- Which category appears most common? **[Write here]**
- Is any information missing? **[Write here]**
- Approximately how many participants are in the smallest observed category? **[Write here]**

### Your turn: copy, paste, and adapt

Copy the bar-chart code and make a bar chart of `gender`. Add `coord_flip()` after `theme_minimal()` so the longer category labels are readable. Give the graph an accurate title and axis labels.

```{r}
#| eval: false

```

### Your turn: fill in the blanks

Create another relationship-status bar chart.

```{r}
#| eval: false
ggplot(data = dat, aes(x = _____)) +
  geom_bar() +
  labs(
    title = "_____",
    x = "_____",
    y = "_____"
  ) +
  theme_minimal()
```

## 4. Visualizing a numeric variable

### A histogram

A histogram divides a numeric variable into intervals called **bins** and shows how many observations fall in each interval.

```{r}
ggplot(data = dat, aes(x = social_confidence)) +
  geom_histogram(
    binwidth = 0.5,
    boundary = 1,
    color = "white",
    fill = "steelblue"
  ) +
  scale_x_continuous(breaks = 1:7) +
  labs(
    title = "Distribution of social confidence",
    x = "Social confidence (1–7)",
    y = "Number of participants"
  ) +
  theme_minimal()
```

Look at the histogram before calculating statistics.

- Where is the approximate center of the distribution? **[Write here]**
- Does the distribution appear roughly symmetric or noticeably skewed? **[Write here]**
- What values describe its approximate range? **[Write here]**

Some values are missing. The histogram uses the available observations.

```{r}
sum(is.na(dat$social_confidence))
```

### Your turn: change the variable

Copy the histogram code and make a histogram of `openness`. Openness also uses a 1–7 scale, so most of the existing settings can remain unchanged.

```{r}
#| eval: false

```

### Your turn: explore bin width

Copy the social-confidence histogram and change `binwidth` from `0.5` to `1`. How does the graph change? Neither graph invents different data; the bin width changes how much detail is displayed.

```{r}
#| eval: false

```

**What changed? [Write 1–2 sentences here.]**

## 5. Visualizing relationships

### Two numeric variables: a scatterplot

Each point in a scatterplot represents one observation. Its horizontal and vertical positions show two numeric values for that observation.

```{r}
ggplot(data = dat, aes(x = extraversion, y = social_confidence)) +
  geom_point(alpha = 0.6, color = "steelblue", na.rm = TRUE) +
  labs(
    title = "Extraversion and social confidence",
    x = "Extraversion (1–7)",
    y = "Social confidence (1–7)"
  ) +
  theme_minimal()
```

Describe the **direction**, **form**, and **strength** of the relationship. Also look for unusual observations.

- Is the direction positive, negative, or neither? **[Write here]**
- Does a straight-line pattern seem reasonable? **[Write here]**
- Would you describe the relationship as weak, moderate, or strong? **[Write here]**

This is a descriptive pattern. The graph does not show that one trait causes the other.

### Your turn: transfer the pattern

Create a scatterplot with `dominance` on the horizontal axis and `self_perceived_status` on the vertical axis.

```{r}
#| eval: false
ggplot(data = dat, aes(x = _____, y = _____)) +
  geom_point(alpha = 0.6, color = "darkorange", na.rm = TRUE) +
  labs(
    title = "_____",
    x = "_____",
    y = "_____"
  ) +
  theme_minimal()
```

### One categorical and one numeric variable: a boxplot

A boxplot provides a compact view of the center, spread, and possible unusual values of a numeric variable within categories.

```{r}
ggplot(data = dat, aes(x = relationship_status, y = social_confidence)) +
  geom_boxplot(na.rm = TRUE, fill = "skyblue") +
  labs(
    title = "Social confidence by relationship status",
    x = "Relationship status",
    y = "Social confidence (1–7)"
  ) +
  theme_minimal()
```

Use the graph to compare medians, spreads, and overlap. Do not treat a visible difference as proof of a population difference; inference belongs in Workshop 3.

### Your turn: adapt the boxplot

Copy the boxplot code and display `leadership_interest` by `relationship_status`.

```{r}
#| eval: false

```

## 6. Numerical descriptive statistics

Graphs preserve features of a distribution that a few numbers cannot. Numerical statistics then give us precise summaries of selected features.

### Center, spread, and sample size

```{r}
dat |>
  summarize(
    n = sum(!is.na(social_confidence)),
    mean = mean(social_confidence, na.rm = TRUE),
    median = median(social_confidence, na.rm = TRUE),
    sd = sd(social_confidence, na.rm = TRUE),
    iqr = IQR(social_confidence, na.rm = TRUE),
    minimum = min(social_confidence, na.rm = TRUE),
    maximum = max(social_confidence, na.rm = TRUE)
  )
```

- The **mean** and **median** summarize center.
- The **standard deviation (SD)** and **interquartile range (IQR)** summarize spread in different ways.
- The minimum and maximum describe the observed range.
- `n` records how many nonmissing observations contributed.

The mean and SD are affected more by extreme values and skew than the median and IQR. Always look at the graph before choosing how to describe the distribution.

### Your turn: summarize the graph you made

Calculate the sample size, mean, median, SD, and IQR of `openness`.

```{r}
#| eval: false
dat |>
  summarize(
    n = sum(!is.na(_____)),
    mean = mean(_____, na.rm = TRUE),
    median = median(_____, na.rm = TRUE),
    sd = sd(_____, na.rm = TRUE),
    iqr = IQR(_____, na.rm = TRUE)
  )
```

Compare these numbers with the openness histogram. Do the mean and median agree with your visual estimate of the center? **[Write here]**

### Counts and proportions

For a categorical variable, counts and proportions usually make more sense than means and standard deviations.

```{r}
dat |>
  count(relationship_status) |>
  mutate(percent = 100 * n / sum(n))
```

The percentages include the missing category because the count includes all 250 rows.

### Your turn: categorical summaries

Copy the preceding pipeline and calculate counts and percentages for `gender`.

```{r}
#| eval: false

```

## 7. Grouped summaries and correlation

### Summaries within categories

`group_by()` tells R to calculate subsequent summaries separately for each category.

```{r}
dat |>
  group_by(relationship_status) |>
  summarize(
    n = sum(!is.na(social_confidence)),
    mean_social_confidence = mean(social_confidence, na.rm = TRUE),
    median_social_confidence = median(social_confidence, na.rm = TRUE),
    .groups = "drop"
  )
```

The `NA` row describes participants whose relationship status is missing. Group sizes matter: summaries based on fewer observations are less stable.

### Your turn: modify a grouped summary

Calculate the sample size, mean, and median of `leadership_interest` within each relationship-status category.

```{r}
#| eval: false
dat |>
  group_by(_____) |>
  summarize(
    n = sum(!is.na(_____)),
    mean_leadership = mean(_____, na.rm = TRUE),
    median_leadership = median(_____, na.rm = TRUE),
    .groups = "drop"
  )
```

### A descriptive correlation

A correlation summarizes the direction and strength of a linear relationship between two numeric variables. It ranges from −1 to +1.

```{r}
cor(
  dat$extraversion,
  dat$social_confidence,
  use = "complete.obs"
)
```

- A positive value describes a positive linear relationship.
- A negative value describes a negative linear relationship.
- A value near zero describes little linear relationship.

Correlation does not establish causation, and a correlation can hide patterns that are visible in the scatterplot. Always use the graph and the correlation together.

### Your turn: summarize a second scatterplot

Calculate the correlation between `dominance` and `self_perceived_status`.

```{r}
#| eval: false
cor(
  dat$_____,
  dat$_____,
  use = "complete.obs"
)
```

Does the correlation agree with the direction and strength you saw in the scatterplot? **[Write here]**

## 8. Final activity: a visual and descriptive report

Create a short report that combines graphs, numerical summaries, and prose. The practice chunks remain non-evaluating so unfinished code cannot break the distributed workbook. Run each chunk interactively while working.

### A. Confirm the data

Use `glimpse()` and report the number of rows and columns.

```{r}
#| eval: false
# Write your inspection code here.
```

### B. Describe one numeric variable

Choose one variable:

- `openness`
- `prestige`
- `social_confidence`
- `leadership_interest`

Create a labeled histogram.

```{r}
#| eval: false
# Write your histogram code here.
```

Calculate its nonmissing sample size, mean, median, SD, and IQR.

```{r}
#| eval: false
# Write your summarize() pipeline here.
```

In 2–3 sentences, describe the distribution and connect the numerical summaries to what appears in the graph.

**[Write here.]**

### C. Describe one relationship

Choose one pair:

- `extraversion` and `social_confidence`
- `dominance` and `self_perceived_status`
- `height_cm` and `grip_strength_kg`

Create a labeled scatterplot.

```{r}
#| eval: false
# Write your scatterplot code here.
```

Calculate the descriptive correlation.

```{r}
#| eval: false
# Write your cor() code here.
```

In 2–3 sentences, describe the direction, form, and strength of the relationship. State explicitly that the result is descriptive and does not demonstrate causation.

**[Write here.]**

### D. Describe one categorical variable

Choose `gender` or `relationship_status`. Create a labeled bar chart, then calculate counts and percentages.

```{r}
#| eval: false
# Write your bar-chart code here.
```

```{r}
#| eval: false
# Write your count-and-percentage pipeline here.
```

Write one sentence identifying the most common category and any missing values or very small categories that affect interpretation.

**[Write here.]**

### E. Render

Save with **Windows: Ctrl+S** or **Mac: Command+S**. Render with the **Render** button or **Windows: Ctrl+Shift+K** / **Mac: Shift+Command+K**.

Double click on`workshop2_student.html` that appears in the project folder to view if it doesn't pop up automatically.