Ethical and Social Issues Around Data Collection
Skill 1.BSkill 5.A
Big idea: Data is only as good as how it was collected. Programmers have to protect people's privacy and watch for data that is incomplete, biased, or collected for a different purpose.
- Explain the risks to privacy from collecting and storing personal data on computer systems.
- Explain the importance of recognizing data quality and potential issues when using a data set.
- Identify an appropriate data set to use in order to solve a problem or answer a specific question.
Privacy
Collecting and storing personal data puts privacy at risk: data can leak, be combined with other data to identify people, or be used in ways users didn't agree to. Collect only what you need, protect what you keep, and be clear with users about how it is used.
Data quality and bias
- Algorithmic bias is systematic, repeated error that creates unfair outcomes for a group of users, often because the data used to build the program underrepresents that group.
- Ask how the data was collected. A survey only of people who already use an app says little about people who don't.
- Incomplete or inaccurate data can make a program work incorrectly or inefficiently.
Choosing the right data set
A data set collected to answer one question may be the wrong tool for another. Weekday morning traffic counts can't tell you about weekend nights.
- Computer use puts personal privacy at risk; programmers should safeguard personal data.
- Algorithmic bias: systematic errors that create unfair outcomes for a group.
- Know how a data set was collected and whether that introduces bias.
- Incomplete or inaccurate data causes incorrect or inefficient programs.
- A data set may not fit a question it wasn't collected to answer.
- More data is not automatically better data. Ask who and what is missing.
A face-recognition system works poorly for one age group. What is the most likely cause?
Show answer
Training data that underrepresents that group, producing algorithmic bias.