2020.09.11 Visualizing Under Constraints


While many of the assignments and project specifications for this class will be open-ended, in reality that is not often the case. In industry settings designers, often must work under specific constraints (limits) based on factors such as the user base, company goals, legal obligations, and technical capabilities. For this activity you will work together in a small group to design under a constraint. The goal of this activity is not to produce a final, polished design. Instead, aim to create a general "sketch" of your visualization. You should be able to tell the class what data you've chosen to visualize, how you decided to visualize them (ideally with a rudimentary sketch), what choices you made during the design process, any trade-offs or alternatives you considered, and how the constraint influenced your final design.

Please follow this procedure:

  1. As a group, read the brief and decide which of the two visualization options you will complete.
  2. Take a look at the table of data you will have access to. Note that you aren't given specific values, just the kinds of attributes you can access. This is intentional. Feel free to invent whatever kind of values you think might fit for those attributes. You can even invent new data attributes if you think they are plausible.
  3. At the bottom of the file, examine the constraints table for your group number. Assuming Zoom doesn't tell you, I'll drop by and provide it.
  4. Start ideating on a visualization (or visualizations). Aim for depth rather than covering everything.
  5. For the next 30 or so minutes, work together to flesh out your ideas. Take notes so that you can later use to provide the information mentioned above to the class..

Your brief

It's October, 2028 and an election is on in the USA. Two front-runners, Candidate Smith of the Bull Moose Party (party color goldenrod) and Candidate Thompson of the Whig Party (party color indigo), are closely competing for the presidency. As a part of a media agency, you have been tasked constructing a web-based dashboard which uses visualization to show users either:

Your group can choose which of these cases they would like to pursue.

Election Result Data

Each "row" of data is an update of election results for a specific "precinct" (a part of a state).

Precinct name
State it is in
Total size of precinct
Amount of votes currently counted
Amount of votes left to count
# Votes for Smith
# Votes for Thompson
Prediction for the precinctAlgorithmic prediction of who will win this precinct (e.g. too close to call, clear Thompson win)
Margin of error for prediction

As new precinct results come in, you generate updated state data.

State name
Number of electoral votes for state
Total # of votes currently counted
Total # of votes left to count
# Votes for Smith
# Votes for Thompson
Prediction for the state:Algorithmic prediction of who will win this state (e.g. too close to call, clear Thompson win)
Margin of error for prediction

Finally, you have country-wide data which is updating frequently.

Number of electoral votes won by Thompson
Number of electoral votes won by Smith
Total # of votes currently counted
Total # of votes left to count
Prediction for the countryAlgorithmic prediction of who will win
Margin of error for prediction
Poll Data

Each "row" of data is a poll. Polls can be nation-wide or on a state level. Assume they are coming in regularly and you have data for some but not all states each day. Thus far you've got about 2 months worth.

Poll title:Name of the poll
Date:When the poll was conducted (can be a range)
Rating:Your company's "rating" of poll quality
Sample size:Number of people polled
Sample details:What kind of people were polled (e.g. registered voters, likely voters, adults)
Sample region:Area where the sampling was done (e.g. country-wide, NY voters)
Results:Polling results (e.g. Thompson 44%, Smith 48%
Margin of error:Possible error in results
Adjusted results:Polling results corrected by your proprietary algorithm

You also have access to these data from a computational model for every day of the past two months.

Estimated candidate results:An estimate of the "true" poll results by aggregating polls
Margin of error for estimate:Possible error in estimate
Group 1:You are designing for mobile devices with small screens.
Group 2:Your users are a community of luddites who haven't seen charts before.
Group 3:Your boss has a shareholder meeting. Include many animations and interactions to impress them.
Group 4:Your users have a vision impairment. Do not make elements that are too small.
Group 5:Your users are on a dial-up modem (4 kilobytes per second) and their monitors can only show 16 color EGA.
Group 6:For some reason, your boss wants to make sure that people can print out your dashboard on a black and white printer.


  1. 538
  2. 538 (2)
  3. NYT page
  4. Real Clear Politics
  5. Polling Report