The personas method is said to present information about people of interest for product design. We propose a formal model to understand persona information in terms of factual attributes. Using an analytic model, we show that the expected prevalence rate of persona descriptions decreases rapidly as attributes are added. We then evaluate this expectation empirically. Using six survey datasets ranging from N=268 to N=10307 respondents and two simulated datasets, we determine the prevalence rates of 10000 randomly generated persona-like descriptions per dataset. Consistent with prediction, we observe decreasing prevalence rates as attributes are added. Pearson's r for observed vs. predicted prevalence, transformed to multinormality, ranges r(9998)=0.394 to r(9998)=0.869 in the sampled datasets (all p< 0.001). Because descriptions with many attributes are likely to represent few people, we suggest that personas should be assessed empirically before they are assumed to describe real groups of people.
展开▼
