2017 Data Storytelling Studio @ MIT

What makes data presentations really powerful, is when users are not only passive consumers. I recently found a striking example for that. It´s a New York Times article called “You Draw It: What Got Better or Worse During Obama’s Presidency”. It shows how indicators of key policy issues developed during the Obama years: e.g. national debt, unemployment rate, number of crimes or troops abroad.

Readers are shown the data for the Bush years (2000-2008). They then can get active and draw what they think might be the line for Obama´s time in office (2008-2016).

In a second step, you are presented with the actual data (blue line) and some additional explanations.

I think this presentation is made for a broad general audience, for readers that are willing to test their own perception against the actual facts. Or as the NYT puts it: “See if you’re as smart as you think you are.”

This piece was published after the November election and before the inauguration of Donald Trump. I see two main goals the authors tried to achieve: first of all, it helps readers assess the performance of the Obama administration on the basis of hard facts. Secondly, it confronts readers with their own potential misjudgments.

Was it effective? I personally was very much attracted by this playful and interactive approach towards stats. I am not sure whether I would have read an article on that topic with just static charts.

 

In a recent article leading up to the Super Bowl, the New York Times used visuals from Second Spectrum analytics to highlight the impact of quarterback pressure on performance. The first visual highlights the disparity between completion percentage for quarterbacks when under pressure versus with a clean pocket to throw from. The visual compares this disparity for all thirty two NFL quarterbacks benchmarked against the league average and particularly highlights the Patriots and the Falcons. The Patriots fall from the fourth ranked to the sixteenth ranked team under pressure while the Falcons only drop from second to fourth.

This comparison leads into the next visual which uses new data released by the NFL to create a heat map showing defender traffic around the pocket in games Brady and Ryan both lost. The graphic also shows the average defenders in the pocket per snap benchmarked against the league average and team average. This element of the visual allows readers to better quantify the significance of the heat map.

These visuals aim to show that while pressure significantly effects the performance of all NFL quarterbacks, Tom Brady is particularly susceptible. Thus, a game plan designed to focus on pressuring Brady is the optimal strategy for the Falcons to defeat the Patriots in the Super Bowl. This data presentation was intended for readers who wanted a deeper analysis of the upcoming Super Bowl, and was successful in doing so by using a new data set to draw an intriguing comparison on the games key players.

Over the course of the last two years, I have become much more interested in politics, particularly the election strategy process. For example, there are so many small ways candidates/parties can influence outcomes by targeting their messaging by demographic, or working to increase/decrease voter turnout in certain regions. In learning more about political issues and this process, I’ve turned to fivethirtyeight, as they often have amazing interactive graphics and aggregate data from many different resources in their models, but target an audience with some education about data analysis.

This interactive visualization allows users to adjust voter turnout percentages and political leanings by demographic, and shows the resultant electoral college map, based off of 2012 election data scaled for population changes. When I came across this graphic a few days ago, I thought it was the perfect tool to test my assumptions about certain demographics. I did this with my thoughts on my home state of Michigan, which usually goes blue, but went red in this election. I decreased black voter turnout a bit, and increased the republican leaning and voter turnout for non-college-educated whites, the two factors which are being most attributed to Trump’s winning of Michigan’s electoral votes. This was just enough to flip Michigan to red, and flipped Pennsylvania and Wisconsin as well, both key states that went red in this election.

This interactive is probably most targeted towards people with a good grasp of the electoral college and specific hypotheses they want to test regarding this past election. It does provide a clean interface for learning about the demographics, both in this aggregate method and on a state-by-state basis further down the page. However, I think it would be more effective with some more demographic factors such as age range. The interactive was also published in October before this election outcome, so having a revised section post-election with the updated data on these percentages of voter turnout would create more relevance and make it more accessible to a wider audience.

This Google Trends interactive shows how searches reflect the way people from around the world think about climate change. It displays search volumes in 20 major cities from 2004-2015, for topics such as energy, recycling, oceans, air pollution, and other words and phrases that relate to the environment.

After selecting a topic, the geometric globe rotates as searches from around the world are displayed. The letters in the searches appear one after another, creating a typing effect. Each search covers roughly the size of an entire continent on the globe, making the globe seem like a small, intimate space, instead of an incredibly vast planet.

You can also click on specific cities and learn more about that city’s concerns regarding a certain environmental topic.

According to Simon Rogers, a contributor to the interactive, the goal was to take Google’s enormous amount of data (there are over 3 billion searches a day) and “make those huge numbers meaningful.” The interactive accomplishes this goal by presenting a snapshot of searches and allowing the user to explore specific topics or cities in just enough detail to pique their interest, but not too much to be overwhelming.

This interactive is geared towards casual views who are interested in learning more about attitudes towards climate around the world at a high level. The interactive isn’t targeted at people who are already experts in this domain or people who want to dive deep into data analysis.

Last semester, I studied voter advice applications. More specifically, I dug into the methodologies used by these applications to recommend political candidates based on the users’ political stances. Most applications provided a ranked list of recommended candidates at the completion of the questionnaires, but a website called The Political Compass took a more passive approach.

The Political Compass represents one’s political ideology through coordinates on a two-dimensional scale, with one axis representing the social spectrum and the other representing the economic spectrum.

After taking their questionnaire, I can see where I’m placed on their ideological scale, as well as where candidates and world leaders are placed. I notice my coordinate position is close to those of certain candidates, which means The Political Compass concluded that those candidates and I share similar ideologies.

This implicit conclusion sends the message that I ought to view those candidates more favorably, and that people on similar positions on the scale are similar in ideology. I can imagine this process elicits varying reactions from users—I’d personally feel a bit upset if my coordinate position was close to Hitler’s. Not to mention, these results present a metric for candidate-to-candidate and candidate-to-user comparisons, which can either confirm or contrast one’s preexisting opinions about the political arena.

Applying the liberal/conservative spectrum to the economic/social axes seems to overly simplify the meaning of ideology, but people do seem to view political stances through binary lenses. Including other factors or dimensions would hopefully signal a shift away from a polarizing approach to politics.

In preparation for Super Bowl LI, I was researching the game to understand the dominance of this sporting event and subsequently stumbled across this infographic produced by WalletHub. The infographic is subdivided into multiple sections; Super Bowl LI, Game Day Notes, Media Extravaganza, Ticket Prices, Super Bowl Ads, Pigging Out at Pigskin Parties, Big-Game Betting, and Super Bowl Economics. Below each section is 3-8 different statistics overlaid on an image.

The audience of this data presentation is sports fans, or at the very least those who have some interest in sports. The statistics present are unconnected, and to be best understood, require some sporting knowledge. This is why I believe the targeted audience is those who are interested in sports.

By providing statistics in a variety of categories that pertain to the Super Bowl, the data presentation aims to convey the magnitude of the game and therefore increase interest in the game. By presenting the massive reach of the game in the form of an infographic readers who do not normally watch the Super Bowl may have a new interest and join the bandwagon by viewing the game. The effectiveness of this data presentation in limited by the lack of comparison for the numbers. For example the infographic lists “70 total cameras used by Fox Sports”. Based on the context of this infographic, one will probably assume this to be a large number, but the reader does not know how many cameras are used in a regular season football game, TV show, blockbuster movie, or another sport. If a comparison was provided, readers can better appreciate the statistic and thus better evaluate the reach of the Super Bowl.

 

Source: https://wallethub.com/blog/super-bowl-facts/1589/

Note: Apologies for the low-quality image. I did not realize the quality until I made the post. The link provides a high quality visual.

I’m definitely an avid NBA fan and a lot of the accounts I follow on Twitter are geared toward providing me with information about the league and the teams I like to keep track of. One Twitter account that I found just a few weeks ago, NBA Math, tries to provide an objective assessment of the league by crunching numbers, specifically through their TPA model that takes into account a per possession conversion of the events that occur throughout the course of games.

Their TPA model often provides a two-number evaluation of players and teams that can be broken down into an offensive and defensive score. Viewing these computations in a table ordered by greatest difference can be a great way to visualize most of the information one would want, but makes comparing quite difficult and tedious.

Many of the tweets they send overcome this by displaying the teams on a simple x-y coordinate graph where the axes represent the offensive and defensive scores.

Adjusted defensive and offensive scores for each NBA team since 1 January 2017 VIA NBA Math

While the aesthetics of the display may be bland, presenting it this way is incredibly powerful and allows a Twitter user to be able to make the quick assessments that are typically favored on a social media platform. They are on the site to get quick bursts of information and this type of data platform allows a single tweet to streamline a lot of information at an incredibly rapid rate. Seeing how various teams positions differ can be done with great ease, allowing a user to identify where there team stands and how they might need to improve relative to the rest of the league and the teams that, by popular opinion, are considered elite.

It’s very unclear how effective a model that tries to boil down performance into two numbers can be useful and effective. However, the ability of NBA Math to send out bursts of tweets in this variety can capture the attention of Twitter users easily allowing them to show off this model and try and convey the information they’re gathering about the league.

Link to Tweet: https://twitter.com/NBA_Math/status/829369301899567104

For the past two years, Bloomberg has been publishing interactive data graphics. I first found them in 2015 after reading their post on climate change. Since then, they posted many more, mostly about politics and finance. A more recent one is about Trump’s cabinet nominees, titled “Trump’s $6 Billion Cabinet: Mostly Men, Mostly White and Not Much Government Experience”.

The data presentation takes the form of two tables: one that compares Trump’s cabinet nominations to the previous two presidents (top picture), and one that gives more detailed information about each nominee (bottom picture). The graphics take special notice of the race, gender, government experience, military experience, and wealth of each cabinet member/nominee.

As the title suggests, the goal of the presentation is first to highlight the similarities and differences between Trump’s cabinet and previous cabinet (his cabinet has roughly the same diversity as Bush’s, but far less government experience than both Bush and Obama), and also to educate viewers about the nominees and cabinet members (achievements, past boards, recent news, public stance on issues, wealth). The categories it chooses effectively reveal a stark contrast between past and proposed cabinets, and the extra information in the table along with the related links for each nominee make it a useful tool to become informed about each person. The experience is a little diminished by the weird scrolling behavior.

The graphic is catered to casual viewers – people who maybe do not follow every political announcement but want a brief overview; a lot of the information would be redundant to people who follow politics vigorously, although the tables give a concise summary that may be good for reviewing.

Overall, Bloomberg’s graphic about the cabinet effectively informs viewers about Trump’s new cabinet – how it’s compares to previous ones, who is on it, and what to expect from the members.