data to evidence

Presenting your evidence

Dr Will Stahl-TimminsAssociate Research Fellow – Visualisation of data and information

Presenting Your Evidence

• Data visualisation at the European Centre• Visual Representation• Design and production methods

% Change inpopulation with

‘good health’+ 95% CI 0.0

-0.5

0.51.01.5

Q1 Q2

All urban areas by income deprivation quintileAll urban areas

Q3 Q4 Q5

Distance of dwelling from sea> 50km 20–50km 5–20km 1–5km

0–1km

CoastHealth

mostdeprived

leastdeprived

0.0-0.5

0.51.01.5

(baseline)

55 75 95

ADAS

-cog

MM

SE SIB

othe

rAD

CS-A

DL

DAD PDS

othe

rNPI

othe

rCI

BIC

GD

SCD

RAD

CS-C

GIC

QoLauthor ageslocation

design, size & follow-up

studyquality

cog

0yr 1 2

no. ofcentres

0 10 20 30

baselineMMSE sex

outcome measures usedfunc be glo

55 75 95

ADAS

-cog

MM

SESIB

othe

rAD

CS-A

DL

DAD

PDS

othe

rNPI

othe

rCI

BIC

GD

SCD

RAD

CS-C

GIC

QoL

0yr 1 2 0 10 20 30 cog func be glo

N = 161

Donepezil 1mg N = 42 M FRandCharBlindAnaly

N = 473

M F

M F

RandCharBlindAnaly

M F

M F

RandCharBlindAnaly

Rogers et al.

1998 (B)

Rogers &

1996? Donepezil 3mg N = 40

Donepezil 5mg N = 39Placebo N = 40

Donepezil 5mg N = 154

Placebo N = 162

N = 468


Placebo N = 153

M FM FM F

Rogers et al.

1998 (A)Donepezil 10mg N = 157

M F

Donepezil 10mg N = 158M F

M F

M F

RandCharBlindAnaly

M FM F

RandCharBlindAnaly

M F

M F

RandCharBlindAnaly

M F

M F

RandCharBlindAnaly

N = 818


Placebo N = 274

N = 60

Donepezil 5mg (D)

Placebo (p)

N = 268


Placebo N = 129

N = 431


Placebo N = 217

Burns et al.

1999

Greenberg et al.

2000

Homma et al.

2000

Mohs et al.

2001

Donepezil 10mg N = 273M F

group 1 (p-D-p-p) N=30group 2 (p-p-D-p) N=30

1mg3mg5mg

5mg

10mg

5mg10mg

5mg10mg

GOfER©

Graphical Overview for Evidence Reviews

Investigating the use of information graphics to explain the effects of climate change on health, comparedto textual presentation.

FLOODS AND STORMS

GLOB

AL T

REND

SFL

OOD

CAUS

ESHE

ALTH

IMPA

CTS

CLIMATE CHANGE

STORMS

DEATH & INJURY

CASE STUDY 1: BANGLADESH

ASSUMPTIONS

Global temperature rise

Global Sea level rise

Increase in monsoon rains

Increase in monsoondischarge into rivers

People affected

Flooding depth

2°c

30cm

18%

5%

4.8%

30–90cm

4°c

100cm

33%

10%

57%

90–180cm

If human activity continues to warm global temperatures,countries like Bangladesh are likely to see more flooding.

CASE STUDY 2: USAStudies in industrialised countries indicate that densely populated urban areas are at risk from sea-level rise.

INFECTIOUSDISEASES

TOXIC CON-TAMINATION

MENTALHEALTH

RAINFALL EVAPORATIONSEA LEVEL SURFACERUN-OFF

LOCALTOPOGRAPHY

URBANISATION

190019502005

FUTU

RE C

HANG

ES

FLOODS

The majority of climate scientists agree that human activity is causing temperatures to rise around the world. As these higher temperatures free water that is usually frozen at the poles, sea levels are rising. Increased temeratures also lead to more evaporation of water from seas and lakes. This can lead to increased rainfall and greater numbers of storms, cyclones and extreme weather events.

Coastal regions are more vulnerable to flooding as sea levels rise.

Extreme rainfall can overwhelm rivers and lakes, causing them to flood.

Drowning by storm surge is the major killer in coastal storms.

Global warming and changes in land use (like urbanisation) affect how much water is carried in the air.

Urbanisation can affect how much excess water can be absorbed into the ground.

Sometimes, the shape of the land can make areas vulnerable to flooding.

The number of people living in cities is growing, particularly in low income countries.

= 100m people in towns or cities

Floods are low-probability, high-impact events that can overwhelm physical infrastructure, human resilience and social organisation.

Floods are the most frequent natural weather disaster. This informationgraphic shows some of the causes and health impacts of floods, and

shows how the number and severity of floods may increase in the future.

LATIN AMERICA

SOUTH ASIA

MICRONESIA

BAY OF BENGAL(particularly atrisk from stormsurges)

VULNERABLE PEOPLE

Those living in Low lying places(especially thosewith high density)

One-quarter of the world’s population resides within 100 km distance and 100 m elevation of the coastline.

In the USA, lower-income groups were most affected by Hurricane Katrina in 2005.

Such as children, theinfirm, or those livingin sub-standard housing.

Poorer communities

Those with limitedability to escape

VULNERABLE PLACES

THE NORTHSEA COAST

SEYCHELLES

THE GULFCOAST

THE NILEDELTA

GULF OFGUINEA

Deaths recorded in disaster databases are from drowning and severe injuries.

Improved warnings have decreased mortality from floods and storm surges in the last 30 years; however, the impact of weather disasters in terms of social and health effects is still considerable and is unequally distributed, particularly affecting women.

VENEZUELA

MOZAMBIQUE

CHINA

1999

2000/2001

2003

30,000 DEAD

1,813 DEAD

130m AFFECTED

Particularly inplaces withpoor sanitation:

From storage orfrom chemicalsalready in theenvironment:

Insufficientlyinvestigated,but may include:

Diarrhoealdiseases

Cholera

Cryptosporidiosis

Typhoid fever

Oil

Pesticides

Heavy metals

Hazardouswaste

Post-traumaticstress

Behaviouraldisorders inchildren

Anxiety?

Depression?

LIKELY EFFECTS

2.5–4m belowsea level by 2100

NEW ORLEANS (USA)

1.5–3m belowsea level now

This would mean that a storm surge from a Category 3 hurricane (estimated at 3 to 4 m without waves) could be 6 to 7 m above areas that were heavily populated in 2004.

Mid-range estimate of 48 cm sea level rise by 2100 plussubsidence

AIR QUALITY AND DISEASE

POLL

UTIO

N SO

URCE

S

ENERGY PRODUCTION

URBAN TRANSPORT

OZONE MODELSFuture emissions are, of course, uncertain, and depend on assumptions of population growth, economic development, regulatory actions and energy use. Changes in concentrations of ground-level ozone driven by scenarios of future emissions and/or weather patterns have been projected for Europe and North America:

There are no projections for cities in low- or middle-income countries, despite the heavier pollution burdens in these populations.

PM MODELSEvidence for the health impacts of particulate matter is stronger than that for ozone. However, there are few models of the impact of climate change on pollutantsother than ozone. These tend to emphasise the role of local abatement strategies in determining the future levels of, primarily, particulate matter, and tend to project the probability of air-quality standards being exceeded instead of absolute concentrations; the results vary by region.

Because transboundary transport of pollutants plays a significant role in determining local to regional air quality, changing patterns of atmospheric circulation at the hemispheric to global level are likely to be just as important as regional patterns for future local air quality.

FOREST FIRES

AIR

POLL

UTIO

N (A

P)

As well as producing greenhouse gasses such as carbon dioxide, that lead to global warming, the burning of fossil fuels for energy releases small particles into the air, called particulate matter (PM).

Naturally-occuring forest fires mean that toxic gaseous and particulate air pollutants are released into the atmosphere.

MORTALITY AND MORBIDITY

PARTICULATEMATTER (PM)Many different kinds of combustion, both artificial and natural, can cause particles of solid matter can become suspended in the air. PM is known to affect morbidity and mortality.

Weather at all time scales determines the development, transport, dispersion and deposition of air pollutants, with the passage of fronts, cyclonic and anticyclonic systems and their associated air masses being of particular importance. This information graphic shows

some of the causes and health impacts of air pollution, and shows how both the amount of air pollution, and our exposure to it, may increase in the future.

Ozone generation is affected by:— Bright sunlight— Raised temperatures— Low winds— Atmospheric moisture

The amount of air pollution breathed in by people depends on:— Wind / circulation of air— Topography— Housing characteristics— Activity patterns

PM generation is affected by:— Raised temperatures — Atmospheric moisture

The number of forest fires is affected by:— Raised temperatures— Atmospheric moisture

In urban areas, transport vehicles are the key sources of nitrogen oxides and volatile organic compounds (VOCs) that lead to ground-level ozone. Burning fossil fuels for transport also releases other gasses and particles.

ALLERGICRHINITIS

OTHERDISEASE

BURNS & SMOKEINHALATIONSevere

allergies can limit quality of life.

Particularly affects children.

Chronic obstructive pulmonary disease.

Can affect quality of life, and is increasingly common, particularly in children.

Other Cardio-vascular and respiratory diseases are also caused by air pollution.

forest fires can have direct effects on health.

COPDPNEUMONIA ASTHMA

OTHER TOXICGASSESAs well as ozone, other toxic gasses such as carbon monoxide can have effects on human health.

OZONEOzone is a secondary pollutant formed through photochemical reactions involving nitrogen oxides and volatile organic compounds (VOCs) in the presence of bright sunshine with high temperatures.

CLIMATE CHANGEThe majority of climate scientists agree that human activity is causing temperatures to rise around the world. These higher temperatures can affect weather systems, causing extremely high or extremely low winds. Rising temperatures also affect the amount of water in the atmosphere.

These changes may affect air pollution in two main ways. First, it may mean that the atmospheric conditions are right for more air pollution to form. Secondly, it may change the patterns of air flow, meaning that more people are exposed to this pollution.

FUTU

RE C

HANG

ES

REFERENCE AREATEMP.

INCREASE EMISSIONS EFFECTS

Knowlton et al., 2004

Bell et al., 2007

Andersonet al., 2001

New Yorkarea, USA

50 cities,East USA

England& Wales

1.6 –3.2°C

1.6 –3.2°C

0.9 –2.4°C

mediumincrease

mediumincrease

noincrease

4.5% moredeaths

0.6% moredeaths

ozone +other AP -

assumed changesby 2050s

(all models assume population constant at year 2000 level)

FLOODS AND STORMSFloods are low-probability, high-impact events that can overwhelm physical infrastructure, human resilience and social organisation. Floods are the most frequent natural weather disaster. Floods result from the interaction of rainfall, surface runoff, evaporation, wind, sea level and local topography. In inland areas, flood regimes vary substantially depending on catchment size, topography and climate. Water management practices, urbanisation, intensified land use and forestry can substantially alter the risks of floods. Windstorms are often associated with floods.

Major storm and flood disasters have occurred in the last two decades. In 2003, 130 million people were affected by floods in China. In 1999, 30,000 died from storms followed by floods and landslides in Venezuela. In 2000/2001, 1,813 died in floods in Mozambique. Improved structural and non-structural measures, particularly improved warnings, have decreased mortality from floods and storm surges in the last 30 years; however, the impact of weather disasters in terms of social and health effects is still considerable and is unequally distributed, particularly affecting women. Flood health impacts range from deaths, injuries, infectious diseases and toxic contamination, to mental health problems.

In terms of deaths and populations affected, floods and tropical cyclones have the greatest impact in South Asia and Latin America. Deaths recorded in disaster databases are from drowning and severe injuries. Deaths from unsafe or unhealthy conditions following the extreme event are also a health consequence, but such information is rarely included in disaster statistics. Drowning by storm surge is the major killer in coastal storms where there are large numbers of deaths. An assessment of surges in the past 100 years found that major events were confined to a limited number of regions, with many events occurring in the Bay of Bengal, particularly Bangladesh.

Populations with poor sanitation infrastructure and high burdens of infectious disease often experience increased rates of diarrhoeal diseases after flood events. Increases in cholera, cryptosporidiosis and typhoid fever have been reported in low- and middle-income countries. Flood related increases in diarrhoeal disease have also been reported in India, Brazil and Bangladesh. The floods in Mozambique in 2001 were estimated to have caused over 8,000 additional cases and 447 deaths from diarrhoeal disease in the following months.

The risk of infectious disease following flooding in high income countries is generally low, although increases in respiratory and diarrhoeal diseases have been reported after floods. An important exception was the impact of Hurricanes Katrina and Rita in the USA in 2005, where contamination of water supplies with faecal bacteria led to many cases of diarrhoeal illness and some deaths.

Flooding may lead to contamination of waters with dangerous chemicals, heavy metals or other hazardous substances, from storage or from chemicals already in the environment (e.g., pesticides). Chemical contamination following Hurricane Katrina in the USA included oil spills from refineries and storage tanks, pesticides, metals and hazardous waste. Concentrations of most contaminants were within acceptable short-term levels, except for lead and volatile organic compounds (VOCs) in some areas. There are also health risks associated with long-term contamination of soil and sediment; however, there is little published evidence demonstrating a causal effect of chemical contamination on the pattern of morbidity and mortality following flooding events. Increases in population density and accelerating industrial development in areas subject to natural disasters increase the probability of future disasters and the potential for mass human exposure to hazardous materials released during disasters.

There is increasing evidence of the importance of mental disorders as an impact of disasters. Prolonged impairment resulting from common mental disorders (anxiety and depression) may be considerable. Studies in both low- and high-income countries indicate that the mental-health aspect of flood-related impacts has been insufficiently investigated. A systematic review of post-traumatic stress disorder in high income countries found a small but significant effect following disasters. There is also evidence of medium- to long-term impacts on behavioural disorders in young children.

Vulnerability to weather disasters depends on the attributes of the person at risk (including where they live, age, income, education and disability) and on broader social and environmental factors (level of disaster preparedness, health sector responses and environmental degradation). Poorer communities, particularly slum dwellers, are more likely to live in flood-prone areas. In the USA, lower-income groups were most affected by Hurricane Katrina, and low-income schools had twice the risk of being flooded compared with the reference group.

High-density populations in low-lying coastal regions experience a high health burden from weather disasters, such as settlements along the North Sea coast in north-west Europe, the Seychelles, parts of Micronesia, the Gulf Coast of the USA and Mexico, the Nile Delta, the Gulf of Guinea, and the Bay of Bengal. Environmentally degraded areas are particularly vulnerable to tropical cyclones and coastal flooding under current climate conditions.

Future vulnerability to climate change

The impacts of developmental, climatic and environmental scenarios on population health are important for health-system planning processes. Also, future trends in health are relevant to climate change because the health of populations is an important element of adaptive capacity.

Coastal flooding is projected to result in a large proportional mortality increase under unmitigated emissions; however, this is applied to a low burden of disease, so the aggregate impact is small. The relative risk is projected to increase as much in high- as in low-income countries.

Particularly vulnerable populations and regions are more likely to suffer harm, have less ability to respond to stresses imposed by climate variability and change, and have exhibited limited progress in reducing current vulnerabilities. For example, all persons living in a flood plain are at risk during a flood, but those with lowered ability to escape floodwaters and their consequences (such as children and the infirm, or those living in sub-standard housing) are at higher risk.

Urban populations are growing faster in low-income than in high-income countries. The urban population increased from 220 million in 1900 to 732 million in 1950, and is estimated to have reached 3.2 billion in 2005. In 2005, 74% of the population in more-developed regions was urban, compared with 43% in less-developed regions. Approximately 4.9 billion people are projected to be urban dwellers in 2030, about 60% of the global population, including 81% of the population in more-developed regions and 56% of the population in less-developed regions.

Urbanisation can positively influence population health; for example, by making it easier to provide safe water and improved sanitation. However, rapid and unplanned urbanisation is often associated with adverse health outcomes. Urban slums and squatter settlements are often located in areas subject to landslides, floods and other natural hazards.

Climate change could have a range of adverse effects on some rural populations and regions, including flood and storm damage, loss of cropping land through floods, and a rise in sea level.

One-quarter of the world’s population resides within 100 km distance and 100 m elevation of the coastline, with increases likely over the coming decades. Climate change could affect coastal areas through an accelerated rise in sea level; an intensification of tropical cyclones; changes in wave and storm surge characteristics; and altered precipitation/runoff. These changes could affect human health through coastal flooding and damaged coastal infrastructure; saltwater intrusion into coastal freshwater resources; damage to coastal ecosystems, coral reefs and coastal fisheries; population displacement; amongst others. Although some Small Island States and other low-lying areas are at particular risk, there are few projections of the health impact of climate variability and change.

Densely populated regions in low-lying areas are vulnerable to climate change. In Bangladesh, it is projected that 4.8% of people living in unprotected dryland areas could face inundation by a water depth of 30 to 90 cm based on assumptions of a 2°C temperature increase, a 30 cm increase in sea level, an 18% increase in monsoon precipitation, and a 5% increase in monsoon discharge into major rivers. This could increase to 57% of people based on assumptions of a 4°C temperature increase, a 100 cm increase in sea level, a 33% increase in monsoon precipitation, and a 10% increase in monsoon discharge into major rivers. Some areas could face higher levels of inundation (90 to 180 cm).

Studies in industrialised countries indicate that densely populated urban areas are at risk from sea-level rise. As demonstrated by Hurricane Katrina, areas of New Orleans (USA) and its vicinity are 1.5 to 3 m below sea level. Considering the rate of subsidence and using a mid-range estimate of 480 mm sea-level rise by 2100, it is projected that this region could be 2.5 to 4.0 m or more below mean sea level by 2100, and that a storm surge from a Category 3 hurricane (estimated at 3 to 4 m without waves) could be 6 to 7 m above areas that were heavily populated in 2004.

1) increased resources2) reduced search3) enhanced pattern recognition4) perceptual inference5) perceptual monitoring6) manipulable mediumThomas, J. J. and K. A. Cook (2005). Illuminating the Path: The Research and Development Agenda for Visual Analytics. Available Online at: http://nvac.pnl.gov/agenda.stm, IEEE Computer Society.!

http://nvac.pnl.gov/agenda.stm,

http://nvac.pnl.gov/agenda.stm,

1) increased resources- high bandwidth of sensory information


vision - 12 MB/stouch - 1 MB/shearing, smell & taste - 1 MB/s

Nørretranders, T. (1999). The user illusion: cutting consciousness down to size. Penguin, London, UK.


Nørretranders, T. (1999). The user illusion: cutting consciousness down to size. Penguin, London, UK.

1 3 7 2

1 3 7 2

50px

150px

350px

100px

1 3 7 2

50px

150px

350px

100px

the 1D size element

1 3 7 2

1 3 7 2

area = 2500px2

area = 7500px2

area = 17500px2

area = 5000px2

1 3 7 2

area = 2500px2

area = 7500px2

area = 17500px2

area = 5000px2

the area element

1 3 7 2the count element

1 3 7 2

1 3 7 20

10

1 3 7 20

10red =

green = blue =

2552310

red = green =

blue =

2551770

red = green =

blue =

255680

red = green =

blue =

2552050

red = green =

blue =

25500

red = green =

blue =

2552550

1 3 7 20

10red =

green = blue =

2552310

red = green =

blue =

2551770

red = green =

blue =

255680

red = green =

blue =

2552050

red = green =

blue =

25500

the colour element

1 3 7 24 6 1 2

1 3 7 24 6 1 2

0

10

1 3 7 24 6 1 2

0

10

2 4 2 1

1 3 7 24 6 1 22 4 2 1

0

10

1

2

3

4

http://www.gapminder.org/videos/200-years-that-changed-the-world-bbc/





colourdistinction used when colour is used arbitrarily for distinction only - to distinguish between sets, for examplecolourdata used when numerical values are represented with colour, on some kind of scale (graded or continuous) or colours are recorded.greyscaledistinction when shading is used arbitrarily to distinguish between sets etc. greyscaledata used when numerical values are represented with greyscale shading.patterndistinction hatched areas or dashed linespatterndatasize1d one dimensional sizing of objects (think bar graphs).size2d two dimensional sizing of objects. (including bubble charts and pie charts, which rely on area)size3d three dimensional sizing of objects.position1d 1 meaningful axis. Data represented by position on a line, circle or other shape, from one point (often representing 0) to another. Can be nominal (no particularly significant order numerically, but could be alpabetical); ordinal (list from highest to lowest for example); interval (position represents numerical value) or ratio (like interval, but with a 0 value meaning nothing, eg. not temperature in celcius)position2d information represented using 2 axes, like a scatter plot, line graph or points on a mapposition3d unusual, and clumsy to represent on paper/screen.textposition1d where the position of text has significance.textposition2dtextposition3dtypesizefontweight use of text that appears bolder or italicised.scale when objects are represented to scaleorientationcountshape (including pictograms and logotypes)symbol assumed convention, such as an arrow for directiongrouping grouping elements, for example using an outline or enclosure.highlighting use of border etc to highlight element/elements. (static highlighting)linking physical joining of elements using a line, etc.physicalrepresentation photos, illustrations.3d used where the representation is physically not flat.density where a grouping or overlapping of elements is used to convey information.homunculus proportional distortion


PhD thesis

http://sites.pcmd.ac.uk/infographics/thesis.php

Chapter 4.1 - Design Elements



1. Use standard graph tools

Open Office / Microsoft Excel / Apple Numbers

2. Use specialist data visualisation tools

Title

Able Baker et al. 2010

Dogg's Hamlet 2009

Echo Bazaar 2006

Gender

M F

M F

M F

M F

Ages Sites

?

Features Outcomes

N=50

N=150

N=25

Design

Control N=28

500mg N=22

http://www.tableausoftware.com/



http://www-958.ibm.com/software/data/cognos/manyeyes/



3. Do it by hand

(or consult a professional information designer)

http://www.informationdesignassociation.org.uk

http://www.informationdesignassociation.org.uk/

http://www.informationdesignassociation.org.uk/

http://www.iiid.net/




In Summary:

Simple graph tools are suitable for simple data.

Online tools (like Gapminder) and specialist software (like GOfER, Tableau, Many Eyes) are available for more complex data.

Bespoke presentations (possibly made by information design professionals) will be needed for the most challenging data. Contact through Information Design Association / International Institute for Information Design.


A word of warning

You may need to present raw data alongside visuals

www.ecehh.org

Dr Will Stahl-TimminsTwitter: @will_s_t

Thank you

data to evidence

Documents

fblind n

frandcharblind2000 n

blind1999analy n

f placebo n

fanalymmsecibic placebo

fanaly placebo n

adlblindadascog2001

mgblind1998 b n