Please read and follow these instructions in order to try these past workshops on your own.

Principles of effective data visualization (using R)

Some basic principles

Taken from various sources (see resources below)

Show the (raw) data (e.g. individual data points)
Minimize ‘ink’ and maximize ‘data’
Keep the graph as simple as is appropriate
Make graph as self-explanatory as possible
Emphasize comparisons between values and groups (e.g. same unit)
Use ‘small multiples’ graphs to show differences between groups
Pay attention to the meaning of the colour scheme (e.g. see ColorBrewer)
- And think of colour blind people!
Focus on the story the data is telling

Things to avoid:

Avoid bar plots: they hide the real data and distort the reality of the data
- (unless the data is proportions/percents)
Avoid SEM as error bars: isn’t a good measure of spread
- (use SD or interquartile range instead)
Avoid pie charts: can easily unintentionally distort the data
- (use bar charts with groups side by side)
Avoid 3D charts: they visually distort the data
- (sometimes 3D is needed e.g. with geography)

Resources

Code

fake_data <- data.frame(
    Group = as.factor(c(rep('Treatment', 10), rep('Control', 10))),
    Weight = c(rnorm(10, mean = 50, sd = 4), rnorm(10, mean = 60, sd = 5))
)

library(dplyr)

#> 
#> Attaching package: 'dplyr'

#> The following objects are masked from 'package:stats':
#> 
#>     filter, lag

#> The following objects are masked from 'package:base':
#> 
#>     intersect, setdiff, setequal, union

summary_fake <- fake_data %>% 
    group_by(Group) %>% 
    summarize(mean = mean(Weight),
              se = sqrt(var(Weight) / length(Weight)))

library(ggplot2)
p1 <- ggplot(fake_data, aes(y = Weight, x = Group)) +
    geom_boxplot()

p2 <- ggplot(fake_data, aes(y = Weight, x = Group)) +
    geom_jitter(width = 0.25)

p3 <- ggplot(summary_fake, aes(y = mean, x = Group)) +
    geom_bar(stat = 'identity', width = 0.5)
    
p4 <- ggplot(summary_fake, aes(y = mean, x = Group)) +
    geom_bar(stat = 'identity', width = 0.5) +
    geom_errorbar(aes(ymax = mean + se, ymin = mean - se), width = 0.25)

library(gridExtra)

#> 
#> Attaching package: 'gridExtra'

#> The following object is masked from 'package:dplyr':
#> 
#>     combine

grid.arrange(p1, p2, p3, p4, ncol = 2)

plot of chunk unnamed-chunk-1

library(ggthemes)
ggplot(fake_data, aes(y = Weight, x = Group)) +
    geom_jitter(width = 0.25) +
    geom_errorbar(stat = 'summary', fun.y = 'mean', width = 0.6, 
                  aes(ymax = ..y.., ymin = ..y..)) +
    theme_tufte(base_family = 'sans')

plot of chunk unnamed-chunk-1

ggplot(iris, aes(y = Petal.Length, x = Petal.Width)) +
    geom_point() +
    facet_grid(~ Species, ) +
    theme_tufte(base_family = 'sans') +
    theme(panel.spacing = unit(2.5, 'lines'))

plot of chunk unnamed-chunk-1

Written on March 1, 2017