It also reminded me of this episode of The West Wing, where C.J. and Josh are briefed on how the Mercator projection map that we all use distorts the relative size and position of the continents.

The visualization is not only beautiful, but also clever, informative, and user friendly.  It takes the most popular supplements and plots them from top to bottom in terms of how strong the scientific evidence is for their relative effectiveness in treating certain conditions.  Additionally, viewers can sort the data by indication to see which supplements are backed (or not) by scientific evidence.

The universe of health supplements is indeed a noisy one, but data visualizations such as this one can help make sense of it all.  Good data visualizations can help separate signals from noise by clearly presenting a large amount of data in a simple, relational way.  In this case, you have the most popular supplements ranked by the degree to which scientific evidence backs their claims.  It is pretty easy to see which claims are valid, and which appear to be nothing but empty promises (and a big waste of money).

AT&T did not file this lawsuit because Verizon’s ‘There’s A Map For That’ advertisements are untrue; AT&T sued because Verizon’s ads are true and the truth hurts.

Verizon in an official response to AT&T’s lawsuit that Verizon’s “there’s a map for that” ad campaign is misleading.

The background is that Verizon recently rolled out a campaign that uses a visualization of relational data to compare their 3G coverage to AT&T.  The visualization is pretty stark.  AT&T fired back that the ad was misleading, giving consumers the impression that AT&T doesn’t have ANY wireless coverage in most parts of the country.

Decide for yourself.  FlowingData has a great recap post with all the ads to date.

David McCandless is a “visual journalist” who specializes in visualizing data across numerous subjects. In his latest work for The Guardian’s Data Blog, David visualize a ton of data regarding troop deaths and injuries, size of forces by country, as well as the civilian toll in Afghanistan.

Like all good visualizers, David tells a story with his infographics, putting many issues in perspective (likely creating some “oh, I didn’t realize” moments for readers) through the use of relational data. For example, David compares absolute measures of troop fatalities by country to fatalities as a percentage of total troops deployed. What one sees is that while the US has lost the most troops by far, the Canadians have lost the most troops as a percentage of those they have deployed:

The piece is chock full of infographics like this. One issue I have with David’s analysis comes less from the data (and it’s visualization), and more with his commentary. One infographic depicts the number of troops in Afghanistan by country or organization (e.g. NATO, etc). David includes a bubble for private security contractors (PSC), which I think is great as it is a key statistic that we should be taking into account. However, while his own data shows that only 3,000 of these contractors are armed he makes the comment that “that’s a huge amount of hired guns”. Now David may just be using the common phrase of ‘hired gun’ to refer to all PSCs, but when talking about PSCs such a phrase implies something very specific (i.e. contractors that carry and use weapons in theater–not just logistical support, etc). If he isn’t just using the phrase in the generic sense then the claim is overblown. By his own numbers, armed PSCs only make up a little over 1% of the entire fighting force in Afghanistan (3,000 out of 292,486). I am not sure 1% constitutes a lot of hired guns.

The best part about how David operates is that he provides links to all his data (including what was used for this article) via Google Docs. This is a great practice, one that encourages readers to check his work, look for additional patterns in the data, and ensures that any errors in presentation or interpretation can be brought to light and discussed. (David has altered other infographics based on reader feedback.) I wish more people would adopt the practice.

In any event, be sure to check out David’s blog and other work.

For those that don’t know him, Rosling is something of a superstar at the intersection of public health research and data visualization.  Rosling’s recent work focuses on dispelling the misconception that there is a binary distinction between the developed and developing worlds.  While Rosling is not the first to argue against this dichotomy (and all its political ramifications), his approach is both fascinating and affective.  Rosling utilizes an impressive visualization tool–that he developed and Google bought–to present data on the progress of the developing and developed worlds in a relational manner.  The video below was shot in June of 2009 at a TED-sponsored talk at the US State Department.

Rosling’s presentation drove home two points for me: the importance of challenging our assumptions, the impact of data visualization, and the power of relational data.

1. Too often we take for granted what the facts are around a given issue. What was true yesterday may not be true today; however, what was true yesterday undoubtedly contributes to our mindset today and the prism through which we view today’s issue. How does one combat this characteristic of human perception? The review of data. We should be challenging our assumptions, testing them against meticulously collected data to ensure that what we once that of as conventional wisdom still applies. If not, change it.
2. How data is visualized can have as big an affect on helping us to alter our views. The data Rosling relies on is not new or proprietary (heck, I used some of it when I lectured on the political economy of development). What is innovative and impactful is how we presents the data.  Rosling’s software visualizes the data in a way that is more powerful than simply providing a list of child mortality rates or the percent change over time for each country.  It allows him to tell a story, to weave a narrative which connects the audience to the data in a way a regular bar chart or data table simply could not do.
3. The most powerful way to visualize data is relationally.  Rosling’s visualizations make full use of this principle, presenting country data relative to other countries as well as across time.

When I think of the difference a data-driven world can and will make, it is this–the ability to view the world and make decisions in a fact-based manner.  We can better separate the wheat from chaff, signal from noise.  We can make our mind set match our data set.  And we are just scratching the service.

Click to Enlarge

Tagged: data visualization, infographic, relational data ]]> http://billpetti.com/2009/10/13/relational-odds-done-right/feed/ 0 billpetti http://billpetti.com/2009/10/12/beta-testing-book-of-odds/ http://billpetti.com/2009/10/12/beta-testing-book-of-odds/#comments Mon, 12 Oct 2009 10:45:45 +0000 Bill Petti http://billpetti.com/?p=702 ]]> I just became a beta tester for a really interesting new venture: Book of Odds.  Book of Odds is a website that attempts to collect and organize as many odds statements as possible:

For over three years we have been building what we believe is the missing dictionary, one filled not with words, but with numbers – the odds of everyday life. It contains hundreds of thousands of Odds Statements, from the odds of being the only one to survive a plane crash, to the odds of having a heart attack, to the odds of having ever eaten cold pizza for breakfast. Book of Odds not only allows you to search for those odds that concern or interest you the most, but also to understand probability by comparing the odds of unfamiliar events to others you have personally experienced. Book of Odds was built for you, and we hope you’ll enjoy it.

So far, the site is fascinating.  I completely agree with the idea that to make odds more meaningful we need to compare them to events that are familiar to us.  As mentioned previously, I am a big believer in the relational presentation of data.  Numbers are great–and I think they are important–but sometimes they can be too abstract, and therefore their impact muted to the casual observer.  Context is key, and presenting data in some comparative fashion can often times provide the grounding and context necessary for their impact to be felt.  So how does Book of Odds accomplish this?

Once you find odds you find interesting you have two ways to compare them: 1) events with odds comparable to your selection are automatically generated to the right of your event, or 2) you can search the site by odd rather than event. Searching broadly based on the odds of your event provides a greater opportunity for comparison.

I’ll have more to say about how useful this is and how functional the site is the more I work my way through. I’ll likely hold any additional comments until they move out of beta. Suffice it to say it is an interesting concept. If you are interested in odds and probability and can still sign up to be a tester I would certainly recommend it.

The essence of behavioral economics is that humans fail to meet the assumptions made by economists in important ways.  As a result, economic models of human behavior often fall short in their predictions of what people “should” do.  By understanding the limitations that actual people have in terms of their knowledge, computational ability, and emotions organizations can design products, services, and processes that lead or ‘nudge’ actors towards better choices. The article mentions a few examples, such as using peer pressure (people do not like being outliers, or being left behind by trends) and short circuiting the inertia of past choices (once we make a choice it tends to stick, regardless of how suboptimal it may be), that have been shown to significantly alter behavior.

One theme that I have heard again and again from some critics is that businesses are using strategies that are underhanded or manipulative in order to get customers to spend money in ways that run against their interests.  No doubt, some may use nudge strategies to take advantage of others, but using a behavioral approach does not necessarily imply that a business is simply trying to manipulate customers into making choices that run counter to their interests–in fact, I would argue the opposite is true.

Because none of us are perfectly rational actors (i.e. homo economicus), we all need help from time to time understanding what action or choice is truly in our self-interest–what choice maximizes value for us. What behavioral economics tells us is that this isn’t as straightforward as previously thought. How a choice and/or information is presented to us makes as much of a difference as what is presented. To me, the lesson to be drawn from behavioral economics is not that businesses need to manipulate their customers, but that they need to be more strategic and more precise in how they make their case for their product or service.  Businesses have a tremendous amount of information about how their service is used by a wide range of actors.  This means they are in a unique position to analyze 1) the most efficient and effective ways to use their services 2) the relative use of their service by different types of clients.  Most clients don’t have access to this kind of information and are therefore at the mercy of their supplier for these insights, or frankly they just guess.

Successful businesses take into account what service will be of the greatest value to clients and customers. By providing great value they ensure that clients will stick around, building a long lasting and reciprocal relationship. If businesses manipulate clients into less valuable purchases in exchange for short term sales, both parties will be worse off in the long run.  The key is to align interests; businesses have an interest in recurring revenue and clients have an interest in finding high quality solutions to their problems–when these interests are aligned both parties will be better off.  (The “Shadow of the Future” plays a significant role here.)

However, as behavioral studies have shown, people don’t always make the best choices for themselves.  Why should customers be any different?  How businesses present choices and service options to clients is just as important as the content of those services.   Clients are people, and therefore have the same barriers to overcome in terms of choices that serve their interests.  In order to nudge clients towards optimal services and service utilization, businesses should present solutions to clients in a relational way–e.g. people like you use this version of the service, or use the service at this rate–presenting solutions in this way can help to counteract the inherent inertia of human behavior.

The key to a relational strategy is data and the visualization of that data.  You have to present the right data in a way that tells a compelling story to the client or else they will not be jolted off of their current path.  Looking back on situations where I was nudged it was typically the result of a profound story about my peers told through compelling data.  It is important to remember that data isn’t just important for bean counters–its critical to story tellers.  The better story you tell with your data about how a client’s peers leverage your service the more likely they are to choose the same behavior, see greater value as a result and appreciate your helping them get there, and, most importantly, remain clients in the future.