---
title: "2.3: Data Visualization"
author: "Ellen Bledsoe, Lily McMullen"
format:
  html:
    toc: true
---

```{r}
#| include: false
knitr::opts_chunk$set(echo = TRUE)
```

# A Visualization Primer

## Learning Outcomes

-   Students will be able to explain why data visualization is important.
-   Students will be able to identify different types of data visualizations and describe when each is appropriate.
-   Students will be able to match a dataset's structure to an appropriate visualization type.
-   Students will be able to critique a visualization.

## Why Does Data Visualization Matter?

### Small Groups

As a group, take 5 minutes to compare your notes and come to a group consensus on your top 3 reasons why data visualization matters. Be sure to choose someone to report out!

::: instructor-only
**Instructor Note:** Some strong answers include: patterns in data become visible that are invisible in a table; complex information can be communicated quickly; visualizations help identify outliers and anomalies; they are essential for communicating findings to non-technical audiences; and they support exploratory data analysis before formal testing.
:::

## Types of Visualization

### Visual cues for communicating data

[Here](https://rafalab.dfci.harvard.edu/dsbook-part-1/dataviz/dataviz-principles.html) is a decent overview of some of the core concepts of data visualization.

This website is pretty great and goes into a lot of detail about good practices in data visualization. If this is something that really piques your interest, it is well worth exploring further!

In this course, our main take-away from this website is the use of visual cues to communicate data and which ones are better than others.

![](assets/data_viz_cues.jpg){width="50%"}

### Small group activity

In groups, discuss as many types of visualizations as you can think of. It's okay if you don't know what they are called! Make a quick list or draw them out if you prefer.

[This website](https://www.data-to-viz.com/) is an amazing reference for data visualization methods and when to use what. It also has examples of each type plotted in `ggplot2`.

We will talk more about how to choose the right visualization for your data now and also in the rest of the module.

::: instructor-only
**Instructor Note:** Students commonly name: bar charts, pie charts, line graphs, scatter plots, histograms. Accept any reasonable answer and write them on the board (if you have one).
:::

### Data matching activity

Complete the data matching activity with your group. Your instructor will provide the worksheet.

Spend a few minutes seeing if you can match the data descriptions to the types of data visualizations.

Each “dataset” is pulled from the same overall dataset: measurements from 344 individual penguins from different species and on different islands over 3 years. Some datasets created multiple plots.

#### Data descriptions

A.  Penguin body mass and flipper lengths
B.  Penguin flipper lengths
C.  Penguin species and flipper lengths
D.  Penguins body mass, species, and flipper lengths

::: instructor-only
**Answers:**

1.  A
2.  C
3.  C
4.  B
5.  C
6.  D
:::

## Figure Critique

First, read through [this blog post](https://humansofdata.atlan.com/2019/02/dos-donts-data-visualization/) on the "Dos and Don'ts of Data Visualization"

Now that you have some insights, let's critique these figures below. Yes, these are actual figures in the wild!

For each one, consider:

-   What aspects don't work?
-   What aspects do work?
-   How would you present this data differently?

**Figure 1:**

![](assets/image1.png){width="50%"}

**Figure 2:**

![](assets/image2.jpg){width="50%"}

**Figure 3:**

![](assets/image3.jpg){width="50%"}

**Figure 4:**

![](assets/image4.png){width="50%"}

::: instructor-only
**Instructor Note:** These figures share several problems that are worth talking about:

-   Truncated axes exaggerate differences by not starting the y-axis at zero. The GNH gender gap looks dramatic visually but is only 0.079 on a 0-1 scale.
-   Pie charts make proportion comparison difficult because humans are poor at judging angles. A bar chart is almost always clearer.
-   3D effects are adding visual distortion without adding information. 3D bars and cylinders make it impossible to accurately read values off an axis.
-   Showing too many groups in the same overlapping plot makes it impossible to find meaning from any of them.
-   Bins of unequal length in a histogram.
:::