5.2. Unfair Discrimination and Social Bias

Learning Outcome

Students will be able to discern bias from fairness in finance, medicine, and society in order to prevent incorrect or distorted conclusions.

Sample Tasks

• Identify data that reflects an unwarranted bias.

• Illustrate the reinforcement of human biases in computer models that make predictions in areas such as medical insurance, financial loans, or policing.

• Describe the effects of biased data as it relates to red-lining.

[OhioDoHEducation21]

Our first reading, from Modern Data Science with R [BKH21], gives some examples of how seemingly neutral algorithms can produce biased results

• 8.6 Algorithmic bias

Thinking Question

• What other proxies for race might be present in the data?

Our second reading is Wikipedia’s entry on Redlining.

Reading Question

• Where did the term redlining start?

• What effects of redlining are still present now?

Further Resources

• Class lecture notes from IFT6758 - Data Science: Data Bias and Algorithmic Discrimination

• The 6 most common types of bias when working with data, a 2021 blog post.

• Data Discrimination: celebrate diversity, not averages, a 2022 blog post.

• Types of Bias in Statistics and the Affect Data Bias Has on Your Business, a post.