Objectivity is a Myth: Your Data is Not Objective and Neither Are You*
Data is often treated as an impartial representation of reality--an unbiased delivery mechanism for "ground truth". Data collection, however, is designed by people, whose knowledge and beliefs influence the design decisions they make. How does that impact what we think we know, and how can we adapt our processes to account for it?
Qualitative research methodologies, like ethnography, are often dismissed by engineers as “subjective”. Data collected by computers—whether sensors or servers—is preferred, presumed to be an objective representation of the world. But is that data as objective as it’s touted to be?
Every data collection scheme is designed by people. The decisions of what to measure and how to measure it are made by people. In each data set, there are layers and layers of choices and assumptions made by people.
In order for the data to be objective, the people designing the collection would have to be objective as well. While a perfectly objective person may exist as an aspirational figment, in reality we’re all blinded by our own perspectives.
But if we’re not objective, and our data isn’t objective, how can we know anything about anything at all? How do we adapt out processes to account for the lack of objectivity which we as a culture so revere? How might we benefit by acknowledging and celebrating subjectivity?
This talk will unpack these issues and provide practical suggestions for understanding both your own perspective and the perspectives baked into the data you use.
data, perception, cognition, objectivity, subjectivity
I gave two talks at OSBridge 2014, "Data Wrangling: Getting Started Working with Data for Visualization":http://www.akashiclabs.com/osbridge-talk-open-source-is-not-enough-the-importance-of-algorithm-transparency/ and "Open Source is Not Enough: The Importantce of Algorithm Transparency":http://www.akashiclabs.com/osbridge-talk-open-source-is-not-enough-the-importance-of-algorithm-transparency/
I have also presented at IEEE Vis 2013, and more recently at the PDX Design Research Group. This talk, however, is brand new!
Data scientist. Data visualizer. Ethnographer. Iconoclast. Pragmatist. Champion for reasonableness. Lover of science and kale.
- Title: Dog Food is for Dogs: Escape the Crate of Your Perspective with User Research
- Track: Cooking
- Room: B202/203
- Time: 10:00 – 11:45am
Dogfooding—using your own products—is nice, but is it sufficient to produce good design for people who aren’t you? Our familiarity with our projects and their quirks makes us poor substitutes for users in the wild. So just who are these users, and how do you incorporate them into design and development?
In this workshop, we’ll explore user experience design and research strategies that will help you design for people who aren’t you.
- Speakers: Rachel Shadoan, amelia abreu