Rashmi Sinha: Persona Creation for Information Rich Sites

Christina pointed to Rashmi Sinha's weblog entry Creating personas for information-rich websites, in which Rashmi proposes a methodology for creating personas that utilizes statistical analysis of user needs and suggests that accuracy is in fact important to persona design. The sugggestion about accuracy is contrary to the tenet in Coopers Inmates... that precision is more important than accuracy. From Rashmi's article describing the methodology, this statement seems important to me,

    Personas for information-rich sites must incorporate input about ways in which people will use complex information domains.
The methodology:

    (a) Use survey techniques (also used in market segmentation). (b) Focus questions around user needs rather than what they simply like / dislike. (c) Identify constellations of needs rather than clusters of users. (d) Use this information as the kernel to build personas around
This was a great find for me. I'm currently working with user surveys and usage statistics to describe the use of a digital library (traditional library measurement). I'm also doing more traditional surveys of user needs to create design personas (design methodology). What I was hoping to do was to use the information-use data to inform the persona development. That way I can provide accurate descriptions of users through personas, which I know is not exactly the Cooper way. There's just a great deal of thrust in my organization to be sure that user behavior that is currently high volume (popular) is not dismissed in any redesign of our products and services. I'm sort of gravitating to Rashmi's model now because of something she mentions in this requiremenet of her methodology, which I agree with, and which should help in bolstering support for personas in my organization, should I use this methology:

    The method should help ground the personas in reality (common critique of personas is that they are based on the designer’s imagination).
Great concepts. I just wouldn't want to break out SPSS to do this, though. I hated statistics in grad school. I suppose identifying constellations of needs is simple enough, though.