the value of data visualisation · the value of data visualisation. research write edit design a...
TRANSCRIPT
research write edit design
research write edit design
Sarah Olesen
Presentation to IPAA Canberra Evaluation Forum
22 September 2016
The value of data visualisation
research write edit design
A clip from the ABC’s ‘Checkout’ about the perils of pie
charts and the importance of accuracy for integrity:
http://iview.abc.net.au/programs/checkout/LE1502H0
10S00
*Start at 24:44 minutes
into the show
research write edit design
Confessions of a data addict enthusiast
research write edit design
Open data
DataEvidence-
based policy
Big data
AccessibilityEvidence-
based policy
The data landscape
research write edit design
Trifecta: Engaging, clear, accurate
Presenting data in a visual way to draw readers in and
make messages clear.
Data visualisations include:
• graphs
• tables
• maps
• infographics.
0
10
20
30
40
50
60Outcome 1
Outcome 2
Outcome 3Outcome 4
Outcome 5
2008 2010 2012
It might be visually interesting, but is this the most intuitive way to show change over time?
Be wary that connected lines imply related and continuous measures
2.50
2.60
2.70
2.80
2.90
3.00
2008 2009 2010 2011 2012 2013 2014 2015
Incidence of Condition A
0.00
0.50
1.00
1.50
2.00
2.50
3.00
3.50
2008 2009 2010 2011 2012 2013 2014 2015
Incidence of Condition A
Always start bars from zero
research write edit design
Why visualise?
1. To make information and findings clear(er)
2. To communicate your message, or ‘data
story’
3. To make data meaningful to readers.
research write edit design
When to visualise?
Making findings clear (but still accurate), when
you:
• have a lot of data and want to draw out the
key and/or overall findings
– over time
– over levels or clusters
– across measures.
research write edit design
When to visualise?
Making findings clear (but still accurate), when:
• have a lot of data
• the overall ‘story’ or pattern(s) in the data is
more important than individual data points.
Image from: Marland et al. 2007 (https://commons.wikimedia.org/wiki/File:Global_Carbon_Emissions.svg)
research write edit design
What to visualise?
Finding your data ‘story’
Ask yourself:
• what pattern(s) is the data showing?
• what message(s) do you want to tell your
audience?*
research write edit design
What to visualise?
Finding your data ‘story’
Ask yourself:
• who/what has changed? (who/what hasn’t?)
• who/what is different?
• what did you find that you weren’t expecting?
• what do you want readers to do about it?
• how can I incorporate data in the overall story of the publication?
Figure 1: Expenditure by district
0
2
4
6
8
10
12
14
North South East West
Before After
What’s interesting?
Figure 1: Men and women’s responses to satisfaction survey
020406080100
Men
0 20 40 60 80 100
Question 1
Question 2
Question 3
Question 4
Question 5
Question 6
Question 7
Question 8
Women
research write edit design
Common data patterns to visualise
• Comparing magnitude (one or more groups)
• Change over time
• Part-to-whole
• Distribution (once, or over time)
• Deviation
• Associations
research write edit design
Match patterns to graph types
research write edit design
https://www.sciencestyle.com.au/
research write edit design
Match patterns to graphs
..and/or other visuals, including:
• tables – when individual data points are important
• maps – when data are spatially located and it helps
to see the spatial relationships between data
• data-driven infographics – to summarise, for text-
based data
Example map: Uptake of screening services for condition x, Victorian local
geographic areas
Image from: http://www.chiefscientist.gov.au/2016/07/science-and-maths-in-australian-secondary-schools-datasheet/
http://www.chiefscientist.gov.au/2016/07/science-and-maths-in-australian-secondary-schools-datasheet/
research write edit design
Make it meaningful
Choose the right visual for your audience
• Know your data and data relationship
• Know your audience
• Know your medium
- Digital/web or print?- Interactive or static?- Accessible? - Which devices?
Policy-makers, decision-makers and managers, public..?
research write edit design
Make it meaningful
• Highlight the reader’s location, job, age-group...
• Use interactive data visualisations let readers
personalise their ‘data story’ by focusing on, or
filtering, the data within the visualisation.
research write edit design
Interactive data visualisation
To interact with this visual go to:https://public.tableau.com/views/LGAScreeningprogramexample/Dashboard1?:embed=y&:display_count=yes&:toolbar=no
research write edit design
Function over fashion
• Keep it simple to keep it clear.
• Reduce clutter (don’t ‘double up’ on data).
• Legibility (font, size, direction).
• Think about data:ink (Tufte 1983).
• Avoid rainbows.
• Avoid bells and whistles (3D effects, shadows) .
• Consider accessibility.
research write edit design
research write edit design
research write edit design
Don’ts
• Don’t use visuals for the sake of visuals.
• Don’t be too complex (eg dual axes).
• Don’t be misleading.
research write edit design
Dos
• Do be intuitive in your
choice of graph type.
• Do be clear and
accurate with statistics.
• Do be consistent.
• Less is often more, but
more is often better
than one complex
visual.
• Do use scientific
conventions.
• Always reference your
source (and always
license your own data).
research write edit design
Take away
• Use data visualisations to summarise and show
patterns and key messages.
• Prioritise clarity and accuracy.
• Know your readers – they will expect:
– accurate
– clear comparisons
– clear outcomes (positive or negative)
– bottom-line messages.
research write edit design
Resources
Good books on data visualisation
• Stephen Few’s ‘Show me the numbers’.
• Edward Tufte’s ‘The visual display of quantitative information’.
• S. Evergreen’s ‘Presenting data effectively’.
Check out the Australian manual of scientific style (particularly
sections on ‘Showing’). Find it here:
https://www.sciencestyle.com.au/
research write edit design
research write edit design
www.biotext.com.au
Sarah Olesen
(02) 6282 2280
Contacts details