Having just returned from a trip to London, my interest in the story of Dr. John Snow and the terrible 1854 cholera outbreak in London was rekindled. It's an amazing story. (While in London I kicked around the same streets where the outbreak occurred some 156 years ago in Soho.) The presentation below is a good short talk by Steven Johnston who wrote an entire book on the case called Ghost Map: The Story of London's Most Terrifying Epidemic—and How It Changed Science, Cities, and the Modern World. The case was explored well by Edward Tufte as well in his 1997 book Visual Explanations: Images and Quantities, Evidence and Narrative (pages 27-37) as a good example of how creating a statistical graph could reveal information in a way that leads to discovering the cause of an epidemic. The case is interesting to anyone, not merely epidemiologists, and among many other things, it underscores the usefulness of using visualization — in this case Snow's map — to both help one investigate the problem and show the problem and possible causes to others in a simple way. You can get a quick review and background information of the case here and here. Watch the video below.
Steven Johnson's telling of this story is illustrative and to the point. Yet, as far as the single slide goes, it would have been far more effective had it been larger. It is always such a shame when a presenter has to say something like "You can't really see it here, but if you look...." I like the idea of using just the famous map here as the only visual. But at that relatively small size, the impact of the map was reduced a bit. Always good advice: make your visuals big.
Each black bar represents a death at the residence; deaths cluster near the Broad Street pump. See larger map herein better resolution.
Comments
Any complex visual will not translate into slides. A lot of Tufte's work would need to be reworked if someone wanted to present it in slide format.
Presenting statistics on slides requires simplistic information design.
I work in the litigation consulting sector of presentation design and we recreate these types of visuals often so that the jury can understand what the point or the message of the statistic is.
Sometimes when using a slide alone as the visual it's easy to lose what's most important by preserving all the detail.
The map is effective because it's relatable (although only for those who actually have visited this area). I would get rid of all the side street names, and mark the pump with an icon in one color, and then the deaths in another color (red would work great). Just adding color would instantly make those areas pop out. Only way to use black for deaths would be to use another faded color for the map itself. Even a 50% gray map would automatically de-clutter this visual.
Another alternative would be to use spheres where the shape area is determined by the number of deaths.
When creating a data based visual we always ask, What's the main point we're trying to show here.
It's not the number of individual deaths at each residence, but rather the clusters around the pump, that's what the story is. Then decide what's the best way to show this information.
Any complex visual will not translate into slides. A lot of Tufte's work would need to be reworked if someone wanted to present it in slide format.
Presenting statistics on slides requires simplistic information design.
I work in the litigation consulting sector of presentation design and we recreate these types of visuals often so that the jury can understand what the point or the message of the statistic is.
Sometimes when using a slide alone as the visual it's easy to lose what's most important by preserving all the detail.
The map is effective because it's relatable (although only for those who actually have visited this area). I would get rid of all the side street names, and mark the pump with an icon in one color, and then the deaths in another color (red would work great). Just adding color would instantly make those areas pop out. Only way to use black for deaths would be to use another faded color for the map itself. Even a 50% gray map would automatically de-clutter this visual.
Another alternative would be to use spheres where the shape area is determined by the number of deaths.
When creating a data based visual we always ask, What's the main point we're trying to show here.
It's not the number of individual deaths at each residence, but rather the clusters around the pump, that's what the story is. Then decide what's the best way to show this information.
-Magda Maslowska
Posted by: Hauteslides | July 16, 2010 at 08:36 PM