A figure is worth a thousand words

# A figure is worth a thousand words
## Data visualization in second language research
### Guilherme D. Garcia

#### Newcastle University

<br>

#### University of Southampton, March 2022

---

# Why figures matter

---

# Why figures matter

## Research stages

1. To <span style="color:FireBrick">***explore***</span> our own data
2. To <span style="color:FireBrick">***define***</span> the appropriate analysis
3. To <span style="color:FireBrick">***communicate***</span> our findings effectively

## Rules of thumb

- Be visually clear
- Don't oversimplify the data
- Find the "sweet spot" of information content

]

<img src="garcia_slides_cllear_files/figure-html/unnamed-chunk-1-1.svg" style="display: block; margin: auto;" />
]

---

# Our goal today

.pull-left[
- Go over a simple study on **feedback**
- Explore patterns
- Assess the stats
- Report findings

## Files

- Slides at <a href = "http://guilhermegarcia.github.io" target = "_blank">`guilhermegarcia.github.io`</a>
- We will use some hypothetical data
- Download the data at <a href = "http://osf.io/hpt4g" target = "_blank">`http://osf.io/hpt4g`</a>
- File name: `dataFiles.zip`
- Our data `\(\rightarrow\)` `feedback.csv`

]

.pull-right[
<p style="text-align:center;"><img src="cover.jpeg" alt="Data viz and analysis in SLR" style="width:50%"></p>

]

---

# Our data

## Some data wrangling first

- Once we load our data, we can see it has a **wide format**
- That means a single variable (task item) spans across **multiple columns**
- This is a problem for plotting and analyzing the data

<div style="border: 1px solid #ddd; padding: 0px; overflow-y: scroll; height:200px; overflow-x: scroll; width:900px;  margin-left: auto; margin-right: auto;" class="table"><table>
<caption>Wide format</caption>
 <thead>
  <tr>
   <th style="text-align:left;position: sticky; top:0; background-color: #FFFFFF;"> ID </th>
   <th style="text-align:left;position: sticky; top:0; background-color: #FFFFFF;"> L1 </th>
   <th style="text-align:right;position: sticky; top:0; background-color: #FFFFFF;"> Age </th>
   <th style="text-align:right;position: sticky; top:0; background-color: #FFFFFF;"> AgeExp </th>
   <th style="text-align:left;position: sticky; top:0; background-color: #FFFFFF;"> Sex </th>
   <th style="text-align:left;position: sticky; top:0; background-color: #FFFFFF;"> Feedback </th>
   <th style="text-align:right;position: sticky; top:0; background-color: #FFFFFF;"> Hours </th>
   <th style="text-align:right;position: sticky; top:0; background-color: #FFFFFF;"> task_A1 </th>
   <th style="text-align:right;position: sticky; top:0; background-color: #FFFFFF;"> task_A2 </th>
   <th style="text-align:right;position: sticky; top:0; background-color: #FFFFFF;"> task_A3 </th>
   <th style="text-align:right;position: sticky; top:0; background-color: #FFFFFF;"> task_A4 </th>
   <th style="text-align:right;position: sticky; top:0; background-color: #FFFFFF;"> task_A5 </th>
   <th style="text-align:right;position: sticky; top:0; background-color: #FFFFFF;"> task_B1 </th>
   <th style="text-align:right;position: sticky; top:0; background-color: #FFFFFF;"> task_B2 </th>
   <th style="text-align:right;position: sticky; top:0; background-color: #FFFFFF;"> task_B3 </th>
   <th style="text-align:right;position: sticky; top:0; background-color: #FFFFFF;"> task_B4 </th>
   <th style="text-align:right;position: sticky; top:0; background-color: #FFFFFF;"> task_B5 </th>
  </tr>
 </thead>
<tbody>
  <tr>
   <td style="text-align:left;"> Learner_1 </td>
   <td style="text-align:left;"> Japanese </td>
   <td style="text-align:right;"> 38 </td>
   <td style="text-align:right;"> 9 </td>
   <td style="text-align:left;"> Female </td>
   <td style="text-align:left;"> Explicit correction </td>
   <td style="text-align:right;"> 8.1 </td>
   <td style="text-align:right;"> 71.0 </td>
   <td style="text-align:right;"> 61.5 </td>
   <td style="text-align:right;"> 62.8 </td>
   <td style="text-align:right;"> 77.5 </td>
   <td style="text-align:right;"> 83.5 </td>
   <td style="text-align:right;"> 47.0 </td>
   <td style="text-align:right;"> 72.2 </td>
   <td style="text-align:right;"> 68.2 </td>
   <td style="text-align:right;"> 72.9 </td>
   <td style="text-align:right;"> 78.3 </td>
  </tr>
  <tr>
   <td style="text-align:left;"> Learner_2 </td>
   <td style="text-align:left;"> Japanese </td>
   <td style="text-align:right;"> 32 </td>
   <td style="text-align:right;"> 10 </td>
   <td style="text-align:left;"> Female </td>
   <td style="text-align:left;"> Explicit correction </td>
   <td style="text-align:right;"> 10.6 </td>
   <td style="text-align:right;"> 65.3 </td>
   <td style="text-align:right;"> 77.2 </td>
   <td style="text-align:right;"> 80.4 </td>
   <td style="text-align:right;"> 85.7 </td>
   <td style="text-align:right;"> 80.3 </td>
   <td style="text-align:right;"> 73.1 </td>
   <td style="text-align:right;"> 67.9 </td>
   <td style="text-align:right;"> 70.8 </td>
   <td style="text-align:right;"> 83.3 </td>
   <td style="text-align:right;"> 94.0 </td>
  </tr>
</tbody>
</table></div>

- What we need is a table where each column represents a variable (`tidy format`)
- We can easily accomplish that using the `pivot_longer()` function in R from the <mark>`tidyverse`</mark> package

---

# Our tidy data

```r
dLong = d %>% 
* pivot_longer(names_to = "Task",
*              values_to = "Score",
*              cols = task_A1:task_B5) %>%
  mutate(Task = str_sub(Task, start = -2L, end = -1L ),
         Item = str_sub(Task, start = -1L, end = -1L),
         Task = str_sub(Task, start = -2L, end = -2L)) %>% 
  select(ID:Task, Item, Score)
```

<table>
<caption>Long format (tidy data)</caption>
 <thead>
  <tr>
   <th style="text-align:left;"> ID </th>
   <th style="text-align:left;"> L1 </th>
   <th style="text-align:right;"> Age </th>
   <th style="text-align:right;"> AgeExp </th>
   <th style="text-align:left;"> Sex </th>
   <th style="text-align:left;"> Feedback </th>
   <th style="text-align:right;"> Hours </th>
   <th style="text-align:left;"> Task </th>
   <th style="text-align:left;"> Item </th>
   <th style="text-align:right;"> Score </th>
  </tr>
 </thead>
<tbody>
  <tr>
   <td style="text-align:left;"> Learner_1 </td>
   <td style="text-align:left;"> Japanese </td>
   <td style="text-align:right;"> 38 </td>
   <td style="text-align:right;"> 9 </td>
   <td style="text-align:left;"> Female </td>
   <td style="text-align:left;"> Explicit correction </td>
   <td style="text-align:right;"> 8.1 </td>
   <td style="text-align:left;"> A </td>
   <td style="text-align:left;"> 1 </td>
   <td style="text-align:right;"> 71.0 </td>
  </tr>
  <tr>
   <td style="text-align:left;"> Learner_1 </td>
   <td style="text-align:left;"> Japanese </td>
   <td style="text-align:right;"> 38 </td>
   <td style="text-align:right;"> 9 </td>
   <td style="text-align:left;"> Female </td>
   <td style="text-align:left;"> Explicit correction </td>
   <td style="text-align:right;"> 8.1 </td>
   <td style="text-align:left;"> A </td>
   <td style="text-align:left;"> 2 </td>
   <td style="text-align:right;"> 61.5 </td>
  </tr>
  <tr>
   <td style="text-align:left;"> Learner_1 </td>
   <td style="text-align:left;"> Japanese </td>
   <td style="text-align:right;"> 38 </td>
   <td style="text-align:right;"> 9 </td>
   <td style="text-align:left;"> Female </td>
   <td style="text-align:left;"> Explicit correction </td>
   <td style="text-align:right;"> 8.1 </td>
   <td style="text-align:left;"> A </td>
   <td style="text-align:left;"> 3 </td>
   <td style="text-align:right;"> 62.8 </td>
  </tr>
  <tr>
   <td style="text-align:left;"> Learner_1 </td>
   <td style="text-align:left;"> Japanese </td>
   <td style="text-align:right;"> 38 </td>
   <td style="text-align:right;"> 9 </td>
   <td style="text-align:left;"> Female </td>
   <td style="text-align:left;"> Explicit correction </td>
   <td style="text-align:right;"> 8.1 </td>
   <td style="text-align:left;"> A </td>
   <td style="text-align:left;"> 4 </td>
   <td style="text-align:right;"> 77.5 </td>
  </tr>
  <tr>
   <td style="text-align:left;"> Learner_1 </td>
   <td style="text-align:left;"> Japanese </td>
   <td style="text-align:right;"> 38 </td>
   <td style="text-align:right;"> 9 </td>
   <td style="text-align:left;"> Female </td>
   <td style="text-align:left;"> Explicit correction </td>
   <td style="text-align:right;"> 8.1 </td>
   <td style="text-align:left;"> A </td>
   <td style="text-align:left;"> 5 </td>
   <td style="text-align:right;"> 83.5 </td>
  </tr>
</tbody>
</table>

---

# Data structure
.pull-left[
- 2 tasks, each with 5 items
- Each item has a score

]

]

## Our focus

.pull-left[
- Is there an effect of `[var]` on `Score`?
- `[var]` needs to be theoretically informed
]

<br>

---

# Efficiency in data visualization

---

# Data visualization 😞

- Here's an **inefficient** figure: only one variable is shown (`Feedback`)
- Only averages are displayed; no uncertainty around the mean (error bars)

.pull-left[
<img src="garcia_slides_cllear_files/figure-html/unnamed-chunk-6-1.svg" style="display: block; margin: auto;" />
]

.pull-right[
<img src="garcia_slides_cllear_files/figure-html/unnamed-chunk-7-1.svg" style="display: block; margin: auto;" />
]

---

# Data visualization 😕

- This is better, but notice how `Feedback` looks drastically different now
- Here we're showing the standard errors (SEs) from the means

.pull-left[
<img src="garcia_slides_cllear_files/figure-html/unnamed-chunk-8-1.svg" style="display: block; margin: auto;" />
]

.pull-right[
<img src="garcia_slides_cllear_files/figure-html/unnamed-chunk-9-1.svg" style="display: block; margin: auto;" />
]

---

# Data visualization 🥰

- This is even better, as box plots give us more information
- We now have medians and data dispersion as well as means and SEs

.pull-left[
<img src="garcia_slides_cllear_files/figure-html/unnamed-chunk-10-1.svg" style="display: block; margin: auto;" />
]

.pull-right[
<img src="garcia_slides_cllear_files/figure-html/unnamed-chunk-11-1.svg" style="display: block; margin: auto;" />
]

---

# Data visualization 😍

- Now we also have information on `Task` (no apparent interaction with `Feedback`)
- This allows us to inspect the effects of two variables on `Score`

.pull-left[
<img src="garcia_slides_cllear_files/figure-html/unnamed-chunk-12-1.svg" style="display: block; margin: auto;" />
]

.pull-right[
<img src="garcia_slides_cllear_files/figure-html/unnamed-chunk-13-1.svg" style="display: block; margin: auto;" />
]

---

# Data visualization 🤯

- Now we add information on the `L1` (means): no clear effect (text now hides means and SEs)
- Total variables: `Score ~ Feedback + Task + L1` (all predictors are **categorical** here)

.pull-left[
<img src="garcia_slides_cllear_files/figure-html/unnamed-chunk-14-1.svg" style="display: block; margin: auto;" />
]

.pull-right[
<img src="garcia_slides_cllear_files/figure-html/unnamed-chunk-15-1.svg" style="display: block; margin: auto;" />
]

---

# Data visualization 🤒

- Here we see individual data points (too much information)
- Total variables: `Score ~ Feedback + Task + L1 + ID`

.pull-left[
<img src="garcia_slides_cllear_files/figure-html/unnamed-chunk-16-1.svg" style="display: block; margin: auto;" />
]

.pull-right[
<img src="garcia_slides_cllear_files/figure-html/unnamed-chunk-17-1.svg" style="display: block; margin: auto;" />
]

---

# Which variables to focus on?

---

# Selecting variables

- We can use **classification trees** as *one* method to check which variables matter overall
- `Score ~ L1 + Sex + Hours + Feedback + Task` using `ctree()` from <mark>`party`</mark>

---

# Zooming in on our key variables

- How does `Hours` affect `Score`? More hours `\(\rightarrow\)` better scores
- What if we add `Feedback`? `Recast` seems to be better than `Explicit correction`, but **not always**

.pull-left[
<img src="garcia_slides_cllear_files/figure-html/unnamed-chunk-19-1.svg" style="display: block; margin: auto;" />
]

.pull-right[
<img src="garcia_slides_cllear_files/figure-html/unnamed-chunk-20-1.svg" style="display: block; margin: auto;" />
]

- Now we have a better sense of what to include in our analysis

???

- Notice that points aren't colored 
- That would be too cluttered; no clear visual separation with this data

---

# Statistical analysis

---

# Running our mixed-effects model

- Our tree suggests that `Hours` and `Feedback` matter, so let's focus on them
- Let's run our statistical model now using `lmer()` from <mark>`lme4`</mark>
- Model: **`Score ~ Hours * Feedback + (1 | ID) + (1 | Item)`**

<table style="border-collapse:collapse; border:none;">
<tr>
<th style="border-top: double; text-align:center; font-style:normal; font-weight:bold; padding:0.2cm;  text-align:left; ">&nbsp;</th>
<th colspan="3" style="border-top: double; text-align:center; font-style:normal; font-weight:bold; padding:0.2cm; ">Score</th>
</tr>
<tr>
<td style=" text-align:center; border-bottom:1px solid; font-style:italic; font-weight:normal;  text-align:left; ">Predictors</td>
<td style=" text-align:center; border-bottom:1px solid; font-style:italic; font-weight:normal;  ">Estimates</td>
<td style=" text-align:center; border-bottom:1px solid; font-style:italic; font-weight:normal;  ">CI</td>
<td style=" text-align:center; border-bottom:1px solid; font-style:italic; font-weight:normal;  ">p</td>
</tr>
<tr>
<td style=" padding:0.2cm; text-align:left; vertical-align:top; text-align:left; ">(Intercept)</td>
<td style=" padding:0.2cm; text-align:left; vertical-align:top; text-align:center;  ">60.28</td>
<td style=" padding:0.2cm; text-align:left; vertical-align:top; text-align:center;  ">51.75&nbsp;&ndash;&nbsp;68.80</td>
<td style=" padding:0.2cm; text-align:left; vertical-align:top; text-align:center;  "><strong>&lt;0.001</strong></td>
</tr>
<tr>
<td style=" padding:0.2cm; text-align:left; vertical-align:top; text-align:left; ">Hours</td>
<td style=" padding:0.2cm; text-align:left; vertical-align:top; text-align:center;  ">1.27</td>
<td style=" padding:0.2cm; text-align:left; vertical-align:top; text-align:center;  ">0.79&nbsp;&ndash;&nbsp;1.76</td>
<td style=" padding:0.2cm; text-align:left; vertical-align:top; text-align:center;  "><strong>&lt;0.001</strong></td>
</tr>
<tr>
<td style=" padding:0.2cm; text-align:left; vertical-align:top; text-align:left; ">Feedback [Recast]</td>
<td style=" padding:0.2cm; text-align:left; vertical-align:top; text-align:center;  ">9.28</td>
<td style=" padding:0.2cm; text-align:left; vertical-align:top; text-align:center;  ">2.60&nbsp;&ndash;&nbsp;15.97</td>
<td style=" padding:0.2cm; text-align:left; vertical-align:top; text-align:center;  "><strong>0.007</strong></td>
</tr>
<tr>
<td style=" padding:0.2cm; text-align:left; vertical-align:top; text-align:left; ">Hours * Feedback [Recast]</td>
<td style=" padding:0.2cm; text-align:left; vertical-align:top; text-align:center;  ">&#45;0.65</td>
<td style=" padding:0.2cm; text-align:left; vertical-align:top; text-align:center;  ">&#45;1.29&nbsp;&ndash;&nbsp;-0.02</td>
<td style=" padding:0.2cm; text-align:left; vertical-align:top; text-align:center;  "><strong>0.042</strong></td>
</tr>
<tr>
<td style=" padding:0.2cm; text-align:left; vertical-align:top; text-align:left; padding-top:0.1cm; padding-bottom:0.1cm; border-top:1px solid;">Observations</td>
<td style=" padding:0.2cm; text-align:left; vertical-align:top; padding-top:0.1cm; padding-bottom:0.1cm; text-align:left; border-top:1px solid;" colspan="3">600</td>
</tr>
<tr>
<td style=" padding:0.2cm; text-align:left; vertical-align:top; text-align:left; padding-top:0.1cm; padding-bottom:0.1cm;">Marginal R<sup>2</sup> / Conditional R<sup>2</sup></td>
<td style=" padding:0.2cm; text-align:left; vertical-align:top; padding-top:0.1cm; padding-bottom:0.1cm; text-align:left;" colspan="3">0.075 / 0.725</td>
</tr>

</table>

<br>

- We have significant effects for all our terms, **BUT** <span style = "color:red">**can't**</span> directly compare `Hours` and `Feedback`

---

# Running our mixed-effects model

- Here's the same model with both variables on the same scale
- We can easily rescale variables using `rescale()` from <mark>`arm`</mark>

<table style="border-collapse:collapse; border:none;">
<tr>
<th style="border-top: double; text-align:center; font-style:normal; font-weight:bold; padding:0.2cm;  text-align:left; ">&nbsp;</th>
<th colspan="3" style="border-top: double; text-align:center; font-style:normal; font-weight:bold; padding:0.2cm; ">Score</th>
</tr>
<tr>
<td style=" text-align:center; border-bottom:1px solid; font-style:italic; font-weight:normal;  text-align:left; ">Predictors</td>
<td style=" text-align:center; border-bottom:1px solid; font-style:italic; font-weight:normal;  ">Estimates</td>
<td style=" text-align:center; border-bottom:1px solid; font-style:italic; font-weight:normal;  ">CI</td>
<td style=" text-align:center; border-bottom:1px solid; font-style:italic; font-weight:normal;  ">p</td>
</tr>
<tr>
<td style=" padding:0.2cm; text-align:left; vertical-align:top; text-align:left; ">(Intercept)</td>
<td style=" padding:0.2cm; text-align:left; vertical-align:top; text-align:center;  ">73.42</td>
<td style=" padding:0.2cm; text-align:left; vertical-align:top; text-align:center;  ">66.45&nbsp;&ndash;&nbsp;80.38</td>
<td style=" padding:0.2cm; text-align:left; vertical-align:top; text-align:center;  "><strong>&lt;0.001</strong></td>
</tr>
<tr>
<td style=" padding:0.2cm; text-align:left; vertical-align:top; text-align:left; ">Hours (std)</td>
<td style=" padding:0.2cm; text-align:left; vertical-align:top; text-align:center;  ">6.47</td>
<td style=" padding:0.2cm; text-align:left; vertical-align:top; text-align:center;  ">4.00&nbsp;&ndash;&nbsp;8.95</td>
<td style=" padding:0.2cm; text-align:left; vertical-align:top; text-align:center;  "><strong>&lt;0.001</strong></td>
</tr>
<tr>
<td style=" padding:0.2cm; text-align:left; vertical-align:top; text-align:left; ">Feedback [Recast]</td>
<td style=" padding:0.2cm; text-align:left; vertical-align:top; text-align:center;  ">2.52</td>
<td style=" padding:0.2cm; text-align:left; vertical-align:top; text-align:center;  ">0.94&nbsp;&ndash;&nbsp;4.10</td>
<td style=" padding:0.2cm; text-align:left; vertical-align:top; text-align:center;  "><strong>0.002</strong></td>
</tr>
<tr>
<td style=" padding:0.2cm; text-align:left; vertical-align:top; text-align:left; ">Hours (std) * Feedback<br>[Recast]</td>
<td style=" padding:0.2cm; text-align:left; vertical-align:top; text-align:center;  ">&#45;3.33</td>
<td style=" padding:0.2cm; text-align:left; vertical-align:top; text-align:center;  ">&#45;6.54&nbsp;&ndash;&nbsp;-0.12</td>
<td style=" padding:0.2cm; text-align:left; vertical-align:top; text-align:center;  "><strong>0.042</strong></td>
</tr>
<tr>
<td style=" padding:0.2cm; text-align:left; vertical-align:top; text-align:left; padding-top:0.1cm; padding-bottom:0.1cm; border-top:1px solid;">Observations</td>
<td style=" padding:0.2cm; text-align:left; vertical-align:top; padding-top:0.1cm; padding-bottom:0.1cm; text-align:left; border-top:1px solid;" colspan="3">600</td>
</tr>
<tr>
<td style=" padding:0.2cm; text-align:left; vertical-align:top; text-align:left; padding-top:0.1cm; padding-bottom:0.1cm;">Marginal R<sup>2</sup> / Conditional R<sup>2</sup></td>
<td style=" padding:0.2cm; text-align:left; vertical-align:top; padding-top:0.1cm; padding-bottom:0.1cm; text-align:left;" colspan="3">0.075 / 0.725</td>
</tr>

</table>

<br>

- Now we see that `Hours` actually matters more than `Feedback` type

---

# Visualizing the interaction

- Back to our figure. Notice that each participant has 10 points in the figure (2 tasks `\(\times\)` 5 items)
- It may be better if each point represents the *average* score of each participant across all 10 items

---

# Visualizing the interaction

- Now, each point = grand average of each participant
- Notice how participants' averages are clustered around our trend lines

---

# Visualizing our model output

- It's very easy to plot our model with `plot_model()` from `sjPlot`:

```r
*plot_model(model.std, show.values = T) +
  labs(title = NULL) + theme_classic() + theme(text = element_text(size = 20)) +
  geom_hline(yintercept = 0, linetype = "dashed")
```

---

# Two key figures

- Our data: both variables of interest

<img src="garcia_slides_cllear_files/figure-html/unnamed-chunk-26-1.svg" style="display: block; margin: auto;" />
]

- Our model: estimates and confidence intervals
<br>

<img src="garcia_slides_cllear_files/figure-html/unnamed-chunk-27-1.svg" style="display: block; margin: auto;" />
]

---

# Summary

---

# Take-home message 📝

- Figures help us and readers understand data patterns
  - This is especially true when examining **interactions**
- They also provide an intuitive way to report model estimates

## Dos and Don'ts ❗️

- Maximize efficiency by balancing information and clarity
- Have clear labels and remove anything that is *unnecessary*
- Use colors and shapes to maximize **function** (or improve accessibility with `ggpattern`)
- "Align" figures and statistical analysis: they should complement each other
- Figures should always have **good resolution**

---

# Extras: Bayes; interactive plots

---

# Same model, but now using Bayes

- Here, visualization is **even more important**, since estimates are distributions
- The areas below show the **posterior distribution** of all estimates
- Means and 95% highest density intervals (HDI) are also provided

<br>

---

# Interactive figures

- There are different options in R; <mark>`plotly`</mark> is a great one

- This can be useful **if** interactivity serves a purpose

---

# Thank you!

### Questions 💭

#### `guilhermegarcia.github.io/CLLEAR/garcia_slides_cllear`