Kevin Liang

More UX and Product leaders need to hear Diana DeMarco Brown (she/her)'s wisdom. Diana is a pioneer in the UX field who takes a thoughtful and nuanced approach to assessing the current state of the tech industry. In our recent conversation, Diana shared many insights, including: - How to evaluate a company's culture during the hiring process - Why private equity ownership can result in design divestment - How Product and UX leaders can effectively influence business decisions - Why we don't need to panic with the rise of AI, but we need to think critically and prioritize regulation during this early stage Link to our full conversation: https://lnkd.in/eemDimVb

42

6 Comments

Jeff Sauro and James R. Lewis, could you help clarify why it’s recommended to report the frequency of users who encounter specific issues (e.g., problem_a, problem_b, etc.) in iterative formative usability tests, particularly with small sample sizes (e.g., 4 out of 6 participants experienced problem_a; 2 out of 6 participants experienced problem_b)? From what I understand, using the frequency of issues from a small sample in a single formative usability test isn’t reliable for (1) generalizing the likelihood of occurrence across the broader population, or (2) prioritizing which problems are more “important” based on their frequency in the test. This is especially true since other factors, like severity, can be equally or more important, and as you’ve outlined in your work, severity is independent of how often a problem is encountered. Is this a correct assumption? I’d appreciate any insights on what I might be misunderstanding I understand the binomial probability formula, as explained in Quantifying the User Experience, is helpful for estimating the likelihood of problem detection in terms of minimum probability, likelihood of detection, and the cumulative likelihood of meeting discovery goals. However, I’m stuck on this point: If we set the minimum number of detections to 1, doesn’t the actual probability of occurrence in the broader population for a small sample usability test remain unknown, even if 4 out of 6 participants experienced problem_a? This doesn’t necessarily imply that problem_a occurs more frequently than problem_b, which was experienced by only 2 out of 6 participants, right? I realize the binomial probability model isn’t meant to be used deterministically, but I’m struggling to understand why we report issue frequencies within small formative usability samples when operating iteratively. From my experience, doing so can lead stakeholders to prioritize issues based purely on their frequency in a given round, even when other criteria should be considered Wouldn’t it be more useful to frame it differently, such as saying, “All of these issues are likely to affect 31% or more of users,” assuming we use a probability of .31, aim for an 85% chance of detecting the issue, and set the minimum number of detections to 1? Then we could apply additional evaluation criteria, such as David Travis’ problem severity framework, for prioritization, among other framework Can you help me understand what I’m missing or misunderstanding?

9

6 Comments

Tabs are everywhere, but do you use them properly? Distinguish between types of tabs, design them for visual clarity, and structure their content for usability. #UX #productDesign #usability https://buff.ly/3YyMXuf

7