Read on Twitter
I've been given a large platform this week. I am a cis white woman and I have immense privilege. Since joining NYU seven years ago, I have been working and teaching in the field of survey design. 1/
I found that in my previous statistical training, there was very little investigation into the wording of survey questions, even having written a thesis on measurement. 2/
I was given datasets and treated the values that I got as gold or modifiable on the back-end (e.g. statistically accounting for measurement error) 3/
I think that statisticians have become more aware of issues of measurement error, but few think about the implications of what questions we ask and how we ask them. These implications are not just statistical, but also impact real people and their lives. 4/
When asking questions about identities, we have to be careful about not "othering" (e.g. if you are not a man or a woman, you shouldn't be asked to select "other") and, in my opinion, as much as possible these questions should have the option to self describe. 5/
We also have to be careful about what we are measuring. We know that gender identity and sex assigned at birth are different. What choices are you making about your survey questions and how does that affect the conclusions you will draw from your research? 6/
The importance of understanding the implications of your respondent-facing questions is paramount. When we group people together into categories , we tell them that we only see them in a certain way, rather than complex humans whose identities matter and should be respected. 7/
I've also seen questions that solicit such detailed identity information that isn't necessary for the study and instead de-identifies your sample (e.g. crossing certain variables leaves only one person so their answers are no longer anonymous). 8/
We also need to value work that investigates how the way that we measure identity interacts with and affects how we understand identities, and the implications on people's lives. As statisticians, it's our job to understand the impact of data collection on people. 9/
So I challenge you: the next time you get a dataset, investigate how the data were collected, what was asked, and how that might not only affect your data analysis, but also the people from whom that data was collected. I will work on being better about this too. 10/
