As I was doing some research on personas, I came several different approaches. I have mentioned Steve Mulder before, however there are two approaches that I think should also be required readings before you hend down the persona path.
The traditional standard from Cooper, which starts with user interviews and validated with quantitative data.
"Personas are based primarily on ethnographic user data. Ethnographic techniques are valuable because they assume that an interview subject's attitudes and behaviors are so habitual as to be unconscious."
Getting from Research to Personas and for a bonus check out Cooper's book "The Inmates are Running the Asylum"
And Microsoft Research's answer by Pruitt and Grudinoldto to the expensive up-font costs of Cooper's technique, which starts with a combination of existing quantitative data and follows-up with user interviews.
"We attempt to start each Persona effort from previously executed, large sample market segmentation studies; much like those discussed by Weinstein " "We have yet to start a Persona effort in an area that does not have some existing quantitative and qualitative data. Thus, our own research effort typically comes after we create our [1st pass] Personas.”