Keeper Ines is watering the plant by the front desk when the crew arrives. "Can I share something I have noticed?" she asks.
"Lots of morning members walk right past the screen at the desk. They never look at it."
Comet blinks. "But our plan puts the whole survey on that screen. I thought everybody used it."
Wren writes the word "assumption" on his clipboard. "What do you notice? We guessed about our users without checking."
Nova glows softly. "Would you like a hint?" she asks. "Keeper Ines sees members you rarely see. More points of view catch more mistakes."
Comet turns to you. "Data analyst, how do we fix our guess?"
Computing innovations can reflect human biases through their algorithms or through their data.
Bias can enter at every level of building something. Programmers should take action to reduce it.
One kind of bias is a wrong assumption developers make about their users.
Equity deficits include little exposure to computing and less access to education and training.
Collaboration that includes diverse perspectives helps avoid bias.
The crew only visits in the afternoon, so it never saw the morning members walk past the screen. Keeper Ines did.
In the story, the crew prints paper forms as a backup. On day 3 of the survey week, Wren sees the morning members walk past the screen himself, and the forms go out.
| Made-up guess the crew made | What could go wrong | A possible fix |
|---|---|---|
| Every member reads the front-desk screen | Members who skip the screen are never heard | Offer paper forms too |
| Members visit at the same times the crew does | Morning and weekend members are missed | Ask at four different times |
| Everyone reads the questions the same way | Some answers mean something different | Try the questions on a few members first |
| Week review | True or false? |
|---|---|
| Surveys, interviews, direct observation and user testing are ways to find out what people need. | ? |
| A census asks only some members of a group. | ? |
| Open-box answers may not be uniform. | ? |
| Collecting more data removes bias that comes from the source. | ? |
| Collaboration that includes diverse perspectives helps avoid bias. | ? |
A strong week, data analyst. Tomorrow's Mission Quest turns your family into a tiny survey.