I was at the Accessibility 2.0 conference in London yesterday and heard a couple of comments about personas that indicated they are still not very well understood. Since I have been teaching developers about them for around 10 years, I thought it might be helpful to provide some clarification.
The first thing to realize about personas is that they are primarily roles that users take while interacting with a system. But, instead of referring to users collectively (or to their more abstract role), we create a persona as a tangible representation. This turns out to be a good thing from a psychological perspective since research shows people are more likely to feel positive about an individual than they are a group with similar characteristics. So personas do really help to promote something I call empathetic design (see my paper for the CHI conference on why this is needed).
It important to emphasize that personas must be tangible. (This was one of the inappropriate references I came across.) You cannot on the one hand refer to Bob as a nuclear physicist who enjoys golf but then say he is aged 40 to 49. Bob needs to be a specific age, have a finite number of children and so on. Otherwise he is straddling the boggy ground between persona and user profile.
Returning to the roles issue, does it make sense to have a persona with a disability? While it would help as a reminder that the system being developed needs to be accessible, being disabled is not usually a role – it is an attribute. (The exception would be a system that was designed specifically for a particular disability, like the Nokia Braille Reader.) This does not diminish the importance of accessibility but does highlight the need to consider other contexts of use as a separate dimension. So while personas would provide focus for the tasks performed by defined user communities, the contexts of use would cover issues like user experience, frequency of use, range of abilities and so on. Of course, accessibility should not really be a separate consideration. With few exceptions, all systems should be designed to support users’ preferred or required interaction styles. This is something that most developers knew about when developing for the desktop but seems to have been forgotten in the intermittent gold rush that is Web 2.0.