The first step toward making a solid piece of data visualization is choosing the right type of chart for a project. Quite frequently, data will contain many, many dimensions. It's your job to choose which dimensions to visualize. Similarly, it's your job to choose how to visualize those dimensions.
It's very easy to get sucked into thinking that you need a map for everything. Nowhere is this more true than in the creation of content for elections; everyone thinks of the giant choropleths behind newscasters as they watch results come in, each region turning the hue of the party it voted for. However, maps are primarily beneficial for depicting geographic data and the proximity of things in a physical space. In an American election, Idaho and New Hampshire both have four electoral college seats, but the latter is a fraction of the former's size. New York state is geographically much smaller than Montana, but it has 29 electoral seats to Montana's three. Unless the intent is to show that states in a particular section of the country all vote a particular way, there really isn't much value to displaying that data as a map (beyond the familiarity a map brings, in terms of displaying, an entire country in a recognizable format).
My team at the Financial Times uses a tool called the Visual Vocabulary, which is inspired by Jon Schwabish and Severino Ribecca's Graphic Continuum. To use it, you choose the dimensions of data you want to depict and it suggests several types of charts you can use to do so. Not only that, but the GitHub repo also contains D3 examples of each chart type. To use it, visit ft-interactive.github.io/visual-vocabulary and fork the examples repo at github.com/ft-interactive/visual-vocabulary.
