Exercise: Deceptive Visualization

In this exercise, you will get more experience designing visualizations for a given dataset and practice your skills of deception to gain insight into how visualizations may mislead. You may work in groups of 1-3 people.

Data

You will examine U.S. wages data from 2021, taken from a visualization created by the US Department of Labor’s Women’s Bureau. The original visualization is here.

A CSV file of the data is available under data/us-wages-2021.csv and has been pre-loaded on this page with the variable name wages:

const wages = FileAttachment("../data/us-wages-2021.csv").csv({ typed: true });

Here are the columns of the dataset. Think about the data types (N, O, Q) and whether you might use them as dimensions or measures in your visualizations.

• Sex: Recorded as “Men” or “Women”.
• Race and ethnicity: Reported race according to US Census data (please see the notes from the original visualization for how race was recorded).
• Median Earnings (USD): Median annual earnings for full-time, year-round workers in the US.
• Occupation Group: The types of jobs measured.

Task 1: Create Earnest Visualizations

In this task, you will explore the provided dataset by creating some earnest visualizations that attempt to faithfully convey the data. We’re not trying to fool anyone yet! We’re just trying to understand the data.

Use whatever tools you like to create at least two earnest visualizations. Either implement the visualizations or add screenshot images below.

Earnest Visualization 1

Earnest Visualization 2

Task 2: Create a Deceptive Visualization

Now create at least one deceptive visualization. In class we discussed four types of deceptive visualizations. Consider these visualization types as you design your deceptive visualizations:

• Incorrect Visualizations
• Illegible Visualizations
• Bullshit Visualizations
• Unconventional Visualizations

Deceptive Visualization

Task 3: Reflect on the Consequences

In this task, you will reflect on how the decisions we made in our visualizations (both earnest and deceptive) may help or harm our intended visualization audience, as well as the subjects being analyzed. Please use the ethical considerations “cheat sheet” to guide your team’s discussion. According to these slides, we are the “analysts” creating visualizations about certain “subjects” (here, workers in the US) for a target “audience” (let’s say the general public).

Take notes on your group’s discussion below:

Earnest Visualizations:

• Audience Impact: These visualizations aim to present data transparently and accurately, fostering informed decision-making and building trust. By adhering to ethical standards, they help prevent misinterpretation and ensure the audience receives a truthful representation of the data.

• Subjects Analyzed: Accurate depictions respect the dignity and experiences of the individuals represented, avoiding potential harm that could arise from misrepresentation. This approach aligns with ethical guidelines that emphasize honest data presentation and the avoidance of bias.

Deceptive Visualization:

• Audience Impact: The deceptive visualization, which manipulates the y-axis to exaggerate differences, can mislead the audience, leading to skewed perceptions and potentially flawed conclusions. Such practices undermine trust and can have broader societal implications by spreading misinformation.

• Subjects Analyzed: Misrepresenting data about specific groups can perpetuate stereotypes or biases, causing harm to those communities. It’s crucial to present data ethically to avoid reinforcing negative perceptions or unjust narratives.

Don’t forget to add, commit, and push your exercises to your GitLab repo!