ETC5523: Communicating with Data

Tutorial 2

Author

Michael Lydeamore

Published

1 February 2001

🎯 Objectives

Understand structure of journalism articles
Add data to a story to improve the impact

Introduction

You will be analysing an article from BBC World News about Japan’s ageing population. The article is available online on the BBC website here.

Form small groups (3-5 people) for these exercises.

👥 Exercise 1

In the article, highlight each section of the “inverted pyramid” model. Note that not all parts of the inverted pyramid may be present in the article.

Solution

Main facts: First two paragraphs include the main facts

More detail: Followed by details of why this may be the case.

Quote:: From Japanese Prime MInister Fumio Kishida.

There’s no bio, and little additional detail, likely because this is a very short article and so there was no space to include the extra information.

Is this article a news feature or a news story? Why?

Solution

This is a news story: It is quoting sources and repeating information. There is no inclusion of characters in the story.

Try and write a new headline for the article that conveys a similar message. Remember the principles of concise writing when developing your new headline.

Solution

There are lots of potential approaches to this. Here is one:

Japan’s ageing population: Now the oldest in the world

Exercise 2

With a good understanding of the content of the article, in your groups, you are tasked with finding some data to enrich this story, and producing a graphic to add to the article.

You will need to:

Locate a suitable data source,
Write code to load in the data and manipulate it into a format appropriate for visualisation,
Create a clear, concise graphic that supports the message of the story.

Solution

You could pick a few aspects of the story to look at data from. I’ve chosen the population data, which I’ve sourced from Our World in Data.

library(tidyverse)
library(viridisLite)
library(ggtext)

population_data <- read_csv("population-and-demography.csv")

plot_data <- population_data |>
    # The article mentions Italy, Finland and Japan
    filter(Code %in% c("ITA", "FIN", "JPN")) |>
    select(!Code) |>
    pivot_longer(!c(Entity, Year)) |>
    group_by(Entity, Year) |>
    # The populations of these countries are very different, so we should normalised them to put them on the same scale
    mutate(normalised_value = value / sum(value)) |>
    ungroup() |>
    # We don't want them in alphabetical order: Japan is the takeaway country
    mutate(Entity = factor(Entity, levels = c("Japan", "Italy", "Finland"))) |>
    # Only old populations
    filter(name == "65+ years") |>
    filter(Year >= 1980)

# One trick is to use text labels instead of a legend to draw the reader to a key part of the plot.
# This puts the labels on the right hand side (max(year)) and at the correct height.
labels <- plot_data |>
    group_by(Entity) |>
    summarise(max_year = max(Year), y_pos = normalised_value[which.max(Year)])

colors <- c("#2600ff", "#a38687", "#a38687")

ggplot(plot_data, aes(x=Year, y = normalised_value*100, colour = Entity)) +
    geom_line() +
    geom_point() +
    geom_text(aes(label = Entity, x = max_year+1, y = y_pos*100), data = labels, hjust = 0) +
    labs(x = "Year", y = "Percentage of population", title = "<span style ='color:#2600ff;'>Japan</span> has the world's oldest population") + 
    guides(colour = "none") +
    # Space for labels
    expand_limits(x = 2030) +
    # We are only colouring one line - the one we want people to look at!
    scale_color_manual(values = colors) +
    cowplot::theme_cowplot() +
    theme(plot.title = element_markdown()) +
    annotate(geom = "text", x = 2030, y = -4, label = "Data source: Our World in Data", size = 2, hjust = 1) +
    coord_cartesian(ylim = c(0, 30), clip = "off")

Exercise 3

Present your visualisation to the class. Explain why you chose a particular data source, and what the audience can learn from the visualisation.

Tip

Try to ensure your graphic adds to the story. You want the visualisation to give extra information to the audience.

Solution

This visualisation shows people how the age of Japan’s population has changed over time, which is not otherwise present in the article. It also shows the scale of the difference between Japan and two other European countries.

It could be improved: There are too many years on the x-axis, the y-axis labels should probably have % signs on them, and the title could be better.

For the data source annotation, best practice would be to include a link, although this isn’t generally possible inside ggplot, so we instead include a plain text reference to the website.