2018 Data Storytelling Studio @ MIT

Hubway apparently closes in the winter. This might limit some the seasonality stories with these data 🙁

I think this is my favorite visualization (not because of the topic) but because they have managed to nail down a very specific and powerful narrative: not how many people died, but how many years were not lived. It’s a great example of something that I call “negative space” exploration in a dataset – unpack the storyline so much that in the end you only have one variable (usually a unit of something). The interaction and interface design then help this one variable’s story shine.

Link

This visualization, “Are Democrats/Republicans Winning the Race for Congress?”, shows FiveThirtyEight’s aggregation of generic ballot polls – ie, polls that ask citizens whether they plan to vote for Democrats or Republicans in November. I think their audience is relatively well-educated people with an interest in politics but who don’t work in politics or follow it intensely.  I also think their audience is younger – 20 and 30-somethings – since their content is wells suited to social media. I also think that they tend to live on the coasts/in major metro areas and are left-leaning. The goals of this presentation are to show the change over time in the estimate for the 2018 midterm Congressional election, and also to show the uncertainty inherent in doing polls and poll aggregation. I think that this presentation is effective. The designers used visual hierarchy well so that the first thing users notice is the trendline of the poll aggregation. Leaving the rest of the chart blank until election also underscores the fact that this is the best prediction for election day based on current conditions, and it is likely to change as it has in the past. Showing the 90% confidence interval and highlighting the overlap between the Democratic and Republican confidence intervals in purple helps the audience to understand the imprecision of polls in a way that simply annotating these figures does not. Further, plotting the results of each individual poll allows the audience to quickly identify outliers – this could be very useful to fact-check news outlets that dramatically report outliers without putting them in context. It would be helpful if users could click on an individual poll to see the name and/or see it highlighted in the table underneath, so that you could investigate what’s happening with the outliers or other polls.