Log in or create a free Rosenverse account to watch this video.
Log in Create free account100s of community videos are available to free members. Conference talks are generally available to Gold members.
Summary
Let’s face it, many of us feel daunted by statistics. But we also know that colleagues and clients ask whether our research has “statistically significant” results. Erin’s book Design for Impact helps you to test your hypotheses about improving design, and she guides you through deciding on your effect sizes to help you get those statistically significant results. Caroline’s book Surveys That Work talks about “significance in practice” and she’s not all that convinced about whether it’s worth aiming for statistical significance. Watch this lively session where Erin and Caroline compared and contrasted their ideas and approaches - helped by your questions and contributions.
Key Insights
-
•
Statistical significance often confuses practitioners because it requires mentally flipping hypotheses and disproving nulls, which is cognitively demanding.
-
•
Effect size is critical to understanding whether a change detected by statistics is meaningful in practice, a concept often neglected in statistics education.
-
•
Fast progress isn't necessarily good progress; teams benefit from slowing down and using statistics to ensure they're moving in the right direction.
-
•
Engineers can be reluctant to implement experiments due to the extra coding load, but they respond well when they understand the learning value gained.
-
•
Survey results (the numerical outcomes) are often confused with the number of respondents required for statistical significance, leading to misunderstandings.
-
•
Statistical thresholds like 95% confidence can be adjusted depending on project needs; lower confidence levels are sometimes acceptable.
-
•
A good hypothesis often starts as an intuitive guess, which gets refined over time through repeated testing and data collection.
-
•
Qualitative and quantitative research should be viewed as complementary tools in a holistic research approach rather than opposed methods.
-
•
AI tools can help generate first drafts of survey questions, but human-centered pilot testing is essential to avoid errors and misinterpretations.
-
•
It's common and acceptable to act on results that are significant in practice but not statistically significant, especially when outcomes clearly affect users.
Notable Quotes
"Statistics is hard because you have to flip flop in your head: think of a hypothesis, then a null hypothesis, then try to disprove the null."
"People confuse statistical significance with significance in practice — they want to know if the change is meaningful, not just mathematically significant."
"Fast is not a virtue in and of itself; moving slower and acting with intention ensures you go in the right direction."
"Engineers hate writing more code, so getting them to buy into experiments means showing the value of the learning on the other side."
"You can have an effect size that matters in practice but isn’t statistically significant, like five users failing a key task in usability testing."
"Most science starts with somebody pulling a number out of their ass — it’s okay to start with a gut instinct or guess."
"Statistics is another tool in our toolbox, part of a hierarchy of evidence that includes qualitative and quantitative methods."
"AI can create first drafts of survey questions, but unless you pilot test with real humans, you won’t know if your audience gets it."
"A lot of people think 95% confidence is the only way, but you can adjust confidence levels based on your situation and needs."
"Start with basics like means, minimums, and ranges — statistics rapidly becomes less mysterious and more useful with practice."
Or choose a question:
More Videos
"Timely interactions that log users out without saving progress cause abandonment, especially for people with disabilities."
Sam ProulxOnline Shopping: Designing an Accessible Experience
June 7, 2023
"The system encourages growth not just in craft but also in client engagement and team collaboration."
Ignacio MartinezFair and Effective Designer Evaluation
September 25, 2024
"Design pairs gave strength in numbers, faster turnarounds, and more impact especially when paired with product managers."
Sarah Kinkade Mariana Ortiz-ReyesDesign Management Models in the Face of Transformation
June 8, 2022
"We as humans value warmth information in others more than competence information."
Daniel GloydWarming the User Experience: Lessons from America's first and most radical human-centered designers
May 9, 2024
"Structure beats free form interaction—systematic query frameworks are essential to avoid pitfalls."
Patrick BoehlerFishing for Real Needs: Reimagining Journalism Needs with AI
June 10, 2025
"UX research teams spend significant time manually analyzing interviews, limiting capacity to focus on higher-value tasks."
Andy Barraclough Betsy NelsonFrom Costly Complexity to Efficient Insights: Why UX Teams Are Switching To Voxpopme
September 23, 2024
"Technology moves faster than policymaking — that’s the pacing problem we need to address."
Alexandra SchmidtWhy Ethics Can't Save Tech
November 18, 2022
"A lot of DevOps is about empathy — doing our piece but thinking about the larger connection."
Louis RosenfeldDiscussion: What Operations can teach DesignOps
November 6, 2017
"Carbon is just the starting point for what can be imagined. We're hired for our expertise to adapt, not just follow directions."
Mitchell BernsteinOrganizing Chaos: How IBM is Defining Design Systems with Sketch for an Ever-Changing AI Landscape
September 29, 2021
