Intersectional data use

Intersectional data use

Quick definition: intersections of identity and oppression (race, gender, sexuality, region, nationality, etc.)

Question: “Intersectional data” problematizes the concept of data. What about codes (queer code, women code) - are they machine-accessible?

When data is collected, inevitably ethnicity is always missing (e.g. FB changed to accommodate gender, but not race - there are algorithms that map out race based on last names)

On FB you can check off your political leaning but not your ethnicity

In Turkey, if you look at the most at-risk populations, many cannot access the internet Transparency differs for different groups. Different groups have different needs

“If you can address the needs of women of color, you have fixed all of society.” Individual pleading is a very inefficient way of dealing with this problem.

Different language versions of the internet have different qualities

Intersectional data is more complete - for example, in the 70s the FDA only tested on male rats, which left off important data

“Messy data is good data”

Surveillance and data - do we want to get counted? How do we get people to consent to be counted? Do we only want to rely on data of people who are comfortable with giving consent.

Quoted Study: of college students surveyed at one school, black women were the most harassed

Important: who are the people asking for the data? NYPD? Transparent third party? There are limitations based on labeling, etc. (example Women, Action, and the Media is viewed by mostly women - limited outreach).

Problem: If we take government-funding, our datasets have to be open and accessible for other potentially nefarious purposes

Story: Iceland mapped the genomes of everyone on the island even though not everyone gave their consent (everyone is related) - raises questions about consent and privacy Name extrapolation in India - last names are often caste-related, which is inherently problematic. However, asking people to change their last names is also problematic

Also in India, dating websites “wedding sites” are all divided across caste

Observation: “data” has many meanings. Even with companies existing that have data, at some point we still need a way to mesh data so that those who need that data for their own voices is there.

Intersectional data can also be means of showing how the construction of data can impact marginalized communities by erasing them.

Bruce: Data is not neutral. So we have to get our own data. That’s a lesson, this is data we’re missing. You get what you measure. Data also both sorts and limits. If we decide categories, we exclude those not in those datasets. We all have to look at what is missing.

Action Items

• case file management, use to create metrics or tweet ids. Link social media acounts, and link from Facebook to Twitter. Randi Harper.
• Talk about gathering your own data, and if anybody who does work with data have any best practices.
• Articulate intersectional data might look like. Whitepaper, people who want to write a memo about intersectional data.
• Stats for us, not for them. Best practices for gathering data and how to evaluate another project will use your data

Misogyny Alert - service in India to provide victims with a network of emotional support. Hashtag idea where your friends get to help - stolen by HeartMob