Exploratory Visual Analysis with Tableau

Tableau is a powerful package for constructing visualizations, and is especially suited for conducting exploratory analysis because it allows you to rapidly generate and iterate through the space of data hypotheses and visualization designs. We recommend using it for A2.

In today’s activity, we will learn and practice using Tableau to analyze a dataset on election spending. The Federal Election Commission (FEC) collects a lot of data on campaign finance including how much money candidates and political action committees (PACs) raise, how much they spend, and what they spend it on.

For this activity, we’ll be exploring independent expenditures—the amount spent by “third-parties” (i.e., not the candidates or committees affiliated with candidates) advocating for supporting or opposing a specific candidate—during the last midterm cycle (2021–2022).

The FEC provides this dataset as a comma-separated value (CSV) file, which can be downloaded here. To understand what this data represents, we recommend perusing the corresponding data dictionary.

A recording of this lecture can be found here.

Table of contents

Warm Up Activity (~15 minutes)

This is a large dataset, with many rows and columns. To familiarize ourselves with the data and Tableau, let’s first warm up by constructing the following visualizations.

Total Spent on behalf of the Two Major Parties

And then, broken out by candidate office (note, sort order of bars):

Spenders on behalf of the Two Major Parties

(Note, the specific layout of the circles may differ)

And then, minor spenders (as defined by spending no more than $10k over the midterm elections).

Positive or Negative Campaigning over Time by Party

Note: the logarithmic y-axis scale.

Explore a Question (Remaining Time)

Now that we’re a little bit more familiar with Tableau and this dataset, let’s dive deeper and explore some potentially interesting insights this dataset has to offer. Here are some potential places to start, though you are welcome to follow your own interests:

  • How did the amount of advocacy in favor or in opposition vary between candidates?
    • Which candidates saw the most/least spent in favor of or in opposition to their candidacy?
    • Which candidates had the greatest difference between these amounts?

  • Which states saw the most positive or negative campaigning, and how did this vary over time?

  • What mix of media (e.g., TV, online, content creators, emails, etc.) is used to advocate for candidates?
    • Was there a difference in media between communication about Democratic vs. Republican candidates? Or between House and Senate candidates?
    • How did this mix of media shift over time?

A Shared Document of Our Results

For each visualization that gives you an interesting insight:

  1. Take a screenshot (or use the Worksheet > Copy > Image… menu item).

  2. Add a slide to this slide deck and, in the presenter notes, write a 1-sentence caption describing what you learned about the data quality from the visualization.

  3. Remember to sign your captions with your names and your Kerberos usernames.

As you explore the dataset, you’re likely to encounter other data quality issues—be sure to screenshot those and add them to the deck as well (with a brief caption!).