Exit polls are taken for sample groups of voters leaving polling places after elections. Participants are given a series of questions regarding demographics, voting behavior, and opinions of national issues. Participants have the option of disclosing which candidate or party they voted for in senate races, house races, ballot measures, and the presidential race.
This lets news outlets like CNN communicate information like:
X% of voters in Utah were Male
and
X% of Male voters in Utah voted for [Candidate]
To scaffold the visuals for these exit polls, I took two general approaches.
First, plot the polls as tables. The information is strictly 2-dimensional, and tables provide for easy parsing and scraping of data. Initial designs were populated with 2012 election data.
Fig. 1 - Table View Iteration 1
Fig. 2 - Table View Iteration 2
Fig. 3 - Table View Iteration 3
The second approach plots the polls as horizontal bars. While this is slightly less readable at the details level, it provides a better view of the shape of the answers at a poll level.
Fig. 4 - Horizontal Bar View Iteration 1
Fig. 5 - Horizontal Bar View Iteration 2
To facilitate browsing the many sets of exit polls available, polls were grouped and tagged by top keywords including gender, race, opinion, and economy. These custom tags extended basic keyword extraction on the poll question and answer set.
Fig. 6 - Filtering across polls (click to play)
An autocomplete field prioritizes the custom tags, and after typing will search through the full set of polls for a given race and return relevant results. The question "Is there a correlation between household income and voting ideology?" becomes relatively simple to answer, as the search results persist at the top of the page when navigating across different polls.