Uncertainty

# Uncertainty

**Session 6**

]

---

# Plan for today

.box-2.medium.sp-after-half[Communicating uncertainty]

.box-4.medium.sp-after-half[Visualizing uncertainty]

---

name: communicating
class: center middle section-title section-title-2 animated fadeIn

# Communicating uncertainty

---

---

# The Bay of Pigs

.pull-left-wide[
<figure>
 <img src="img/06/bay_of_pigs.jpg" alt="Planning the Bay of Pigs invasion" title="Planning the Bay of Pigs invasion" width="100%">
</figure>
]

???

When asked by President John F. Kennedy to assess the CIA invasion plan, the U.S. Joint Chiefs of Staff responded that it had a “fair chance” of success. Kennedy took that as a positive assessment. Instead, the Chiefs meant that they judged the chances of success as “3 to 1 against.” But this was never clarified at the time.

Source of story: <https://www.cia.gov/library/center-for-the-study-of-intelligence/csi-publications/books-and-monographs/sherman-kent-and-the-board-of-national-estimates-collected-essays/6words.html>

Chart of perceptions and probabilities: <https://www.cia.gov/library/center-for-the-study-of-intelligence/csi-publications/books-and-monographs/psychology-of-intelligence-analysis/fig18.gif/image.gif>

---

# Misperceptions of probability

.box-inv-2.medium.sp-after[1 in 5 vs. 20%]

]

---

# Misperceptions of probability

.center[
<figure>
 <img src="img/06/forecast-ut.png" alt="UT Senate 2018 forecast from FiveThirtyEight" title="UT Senate 2018 forecast from FiveThirtyEight" width="90%">
</figure>
]

???

UT Senate 2018 election forecast from FiveThirtyEight

---

# Misperceptions of probability

.center[
<figure>
 <img src="img/06/forecast-tx.png" alt="TX Senate 2018 forecast from FiveThirtyEight" title="TX Senate 2018 forecast from FiveThirtyEight" width="90%">
</figure>
]

???

TX Senate 2018 election forecast from FiveThirtyEight

---

# Misperceptions of probability

.box-inv-2.sp-after[Chance of rain = Probability × Area]

.pull-left-wide[
<figure>
 <img src="img/06/rain@4x.png" alt="ZZZ" title="ZZZ" width="100%">
</figure>
]

]

???

When it doesn't rain, it doesn't mean the forecast was wrong!

30% chance doesn't mean it’ll rain just 30% hard. Average over tons of simulations. Depends on geographic area

---

# Misperceptions of probability

.center[
<figure>
 <img src="img/06/coins.jpg" alt="Stack of coins" title="Stack of coins" width="70%">
</figure>
]

???

[Coins](http://www.freestockphotos.biz/stockphoto/8215)

---

# Misperceptions of probability

.pull-left[
<figure>
 <img src="img/06/maria2.png" alt="Hurricane Maria NOAA" title="Hurricane Maria NOAA" width="100%">
 <figcaption>Hurricane Maria map, NOAA</figcaption>
</figure>
]

.pull-right[
<figure>
 <img src="img/06/maria1.png" alt="Hurricane Maria NYT" title="Hurricane Maria NYT" width="100%">
 <figcaption>Hurricane Maria map, New York Times</figcaption>
</figure>
]

???

Or just use a sharpie!

---

# The needle

.center[
<figure>
 <img src="img/06/needle.gif" alt="2016 election needle" title="2016 election needle" width="50%">
</figure>
]

???

Extra data is implicitly encoded in aesthetics - temporal sequence of ballot counting – how did Amanda Cox defend this?

Via https://www.vis4.net/blog/2016/11/jittery-gauges-election-forecast/

---

# The needle

.pull-left[
<figure>
 <img src="img/06/needle-ga.png" alt="Needle GA-7 2017" title="Needle GA-7 2017" width="100%">
</figure>
]

.pull-right[
<figure>
 <img src="img/06/needle-tweets.png" alt="Twitter reactions to needle" title="Twitter reactions to needle" width="100%">
</figure>
]

---

layout: false
name: visualizing
class: center middle section-title section-title-4 animated fadeIn

# Visualizing uncertainty

---

---

# Problems with single numbers

]

]

---

# More information is always better

.box-inv-4.medium[Avoid visualizing single numbers when you have a whole range or distribution of numbers]

---

# Histograms

``` r
library(gapminder)

gapminder_2002 <- gapminder |> 
 filter(year == 2002)

ggplot(gapminder_2002,
       aes(x = lifeExp)) +
  geom_histogram()
```
]

---

# Histograms: Bin width

]

<img src="06-slides_files/figure-html/hist-too-wide-1.png" width="100%" style="display: block; margin: auto;" />
]

<img src="06-slides_files/figure-html/hist-just-right-1.png" width="100%" style="display: block; margin: auto;" />
]

---

# Histogram tips

.box-4.tiny[`geom_histogram(..., color = "white")`]

<img src="06-slides_files/figure-html/hist-border-1.png" width="100%" style="display: block; margin: auto;" />
]

.box-4.tiny[`geom_histogram(..., boundary = 50)`]

]

---

# Density plots

``` r
ggplot(gapminder_2002,
       aes(x = lifeExp)) +
  geom_density(fill = "grey60", 
               color = "grey30")
```
]

---

# Density plots: Kernels and bandwidths

]

<img src="06-slides_files/figure-html/gaussian-bw-10-1.png" width="100%" style="display: block; margin: auto;" />
]

<img src="06-slides_files/figure-html/gaussian-bw-auto-1.png" width="100%" style="display: block; margin: auto;" />
]

---

# Density plots: Kernels and bandwidths

]

<img src="06-slides_files/figure-html/gaussian-kernel-epanechnikov-1.png" width="100%" style="display: block; margin: auto;" />
]

<img src="06-slides_files/figure-html/gaussian-kernel-rectangular-1.png" width="100%" style="display: block; margin: auto;" />
]

---

# Box plots

``` r
ggplot(gapminder_2002,
       aes(x = lifeExp)) +
  geom_boxplot()
```
]

---

# Box plots

---

# Violin plots

.box-4.small[Often helpful to overlay other things on it]

``` r
ggplot(gapminder_2002,
       aes(x = "", 
           y = lifeExp)) +
  geom_violin() +
  geom_boxplot(width = 0.1)
```
]

???

<https://xkcd.com/1967/>

---

# Uncertainty across multiple variables

.box-inv-4.medium.sp-after[Visualize the distribution of a single variable across groups]

.box-inv-4.medium[Add a `fill` aesthetic or use faceting!]

---

# Multiple histograms

``` r
ggplot(gapminder_2002,
       aes(x = lifeExp,
           fill = continent)) +
  geom_histogram(binwidth = 5, 
                 color = "white", 
                 boundary = 50)
```
]

---

# Multiple histograms

``` r
ggplot(gapminder_2002,
       aes(x = lifeExp,
           fill = continent)) +
  geom_histogram(binwidth = 5, 
                 color = "white", 
                 boundary = 50) +
  guides(fill = "none") +
  facet_wrap(vars(continent))
```
]

---

# Pyramid histograms

.left-code.small-code[

``` r
gapminder_intervals <- gapminder |> 
 filter(year == 2002) |> 
 mutate(africa = 
 ifelse(continent == "Africa", 
 "Africa", 
 "Not Africa")) |> 
 mutate(age_buckets = 
 cut(lifeExp, 
 breaks = seq(30, 90, by = 5))) |> 
 group_by(africa, age_buckets) |> 
 summarize(total = n())

ggplot(gapminder_intervals, 
       aes(y = age_buckets,
           x = ifelse(africa == "Africa", 
                      total, -total),
           fill = africa)) +
  geom_col(width = 1, color = "white")
```
]

???

There's no way to use `geom_histogram()` to do this, but we can fake it with `geom_col()`

---

# Multiple densities: Transparency

``` r
ggplot(filter(gapminder_2002, 
              continent != "Oceania"),
       aes(x = lifeExp,
           fill = continent)) +
  geom_density(alpha = 0.5)
```
]

---

# Multiple densities: Ridge plots

``` r
library(ggridges)

ggplot(filter(gapminder_2002, 
              continent != "Oceania"),
       aes(x = lifeExp,
           fill = continent,
           y = continent)) +
  geom_density_ridges()
```
]

---

# Multiple densities: Ridge plots

.center[
<figure>
 <img src="img/06/dwnominate.png" alt="DW-NOMINATE of US House by Party" title="DW-NOMINATE of US House by Party" width="65%">
</figure>
]

---

# Multiple geoms: `gghalves`

``` r
library(gghalves)

ggplot(filter(gapminder_2002, 
              continent != "Oceania"),
       aes(y = lifeExp,
           x = continent,
           color = continent)) +
  geom_half_boxplot(side = "l") +
  geom_half_point(side = "r") 
```
]

---

# Multiple geoms: Raincloud plots

``` r
library(gghalves)

ggplot(filter(gapminder_2002, 
              continent != "Oceania"),
       aes(y = lifeExp,
           x = continent,
           color = continent)) +
  geom_half_point(side = "l", size = 0.3) + 
  geom_half_boxplot(side = "l", width = 0.5, 
                    alpha = 0.3, nudge = 0.1) +
  geom_half_violin(aes(fill = continent), 
                   side = "r") +
  guides(fill = "none", color = "none") +
* coord_flip()
```
]

---

# Uncertainty in model estimates

.center[
<figure>
 <img src="img/06/1-coefs-bayes.png" alt="Dissertation CSRE coefficient plot" title="Dissertation CSRE coefficient plot" width="60%">
</figure>
]

---

# Uncertainty in model estimates

.center[
<figure>
 <img src="img/06/fig-coefs-h3-bayes.png" alt="INGOs and aid coefficient plot" title="INGOs and aid coefficient plot" width="85%">
</figure>
]

---

# Uncertainty in model estimates

.center[
<figure>
 <img src="img/06/results-h1-4.png" alt="Why donors donate halfeye plot" title="Why donors donate halfeye plot" width="75%">
</figure>
]

---

# Uncertainty in model effects

.center[
<figure>
 <img src="img/06/1-ext-pred.png" alt="Dissertation marginal effects plot" title="Dissertation marginal effects plot" width="75%">
</figure>
]

---

# Uncertainty in model outcomes

.center[
<figure>
 <img src="img/06/fivethirtyeight-outcomes.png" alt="FiveThirtyEight model outcomes plot" title="FiveThirtyEight model outcomes plot" width="55%">
 <figcaption>FiveThirtyEight's 2018 midterms model outcomes plot</figcaption
</figure>
]