Words that actually help
I helped run a set of studies asking what the best way is to show sympathy to someone facing stress because they're LGBTQ+, with more than 300 people taking part.
- Role
- Research Assistant
- Methods
- Experimental design, online surveys, qualitative coding, statistical analysis
- Tools
- Qualtrics, Prolific, R, Excel
- Partners
- Professors at Northwestern, the University of Chicago and Yale
- Design the studyProtocols and proposals for ethics review, agreed across four universities
- Build the surveyWritten in Qualtrics, then shared with participants through Prolific
- Collect responsesMore than 300 people, kept anonymous from the first click
- Code and analyzeOpen-ended answers sorted with a shared codebook, numbers run in R
The problem
People who are LGBTQ+ face a steady extra load of stress from how others treat them, which psychologists call minority stress. Friends and family often want to help but don't know what to say. Most relationship research has studied straight couples, partly because queer samples tend to be small, so there was little evidence on which kinds of support land and which fall flat.
When someone shares stress tied to their sexual or gender identity, what response actually helps?
My approach
Ask the people it's about. The main study, and the smaller studies that grew out of it, asked LGBTQ+ participants themselves how different ways of showing sympathy came across. Each study used both kinds of evidence: ratings we could compare in numbers, and open-ended answers in participants' own words.
Handle the data with care. The answers were personal, so participants stayed anonymous and the data was stored to strict standards. Every study went through ethics review first, with proposals written so another team could repeat the work.
Turn words into evidence. We built a codebook so that open-ended answers could be sorted the same way by different people, then analyzed the results in R and Excel.
My part. I worked with the lab team on every stage: designing the study protocols, writing the proposals, building and running the surveys, coding responses and analyzing the data. The professor leading the lab set the direction, and the methods were agreed with collaborators at Northwestern, the University of Chicago and Yale.
What I took from it
Who you ask decides what you learn. Research that leaves out a group by default ends up with nothing to say about it, which is a lesson I now carry into every user study I plan: check who's missing from the sample before trusting the results.