I’ve just read Rachael King’s article in BusinessWeek called “What Your Blog Says About You” . Rather than describe Typealyzer myself, I’ll quote from King’s article:

Typealyzer is a research project that looks at how language reflects a person’s psychological type and his or her motivations and interests. The site was created by Mattias Östmar of PRfekt, a Swedish research and development company focused on media analysis.

For the past two years, Östmar has been collecting sample texts from blogs, based on research about personality type and writing style. The site uses a tool to run a statistical analysis of the text to come up with a word frequency algorithm for different personality types. After the blog is scanned, Typealyzer comes up with personality types derived from the Myers-Briggs model for looking at how people perceive the world and make decisions.

I’m familiar with the Myers-Briggs. In fact, I am qualified to administer and interpret it, and I’ve written about it. (See, for example, my article “Type and System Development: What’s the Connection?” , which is on my personal site and was published in System Development.)

So I ran the Luminanze Blog through Typealyzer, and it says I’m an INTP. Hmmm…

That’s pretty close, actually. Pretty dadgum close. But it’s not quite right: I’m an INFP . I do have strong T skills, though (having always been interested in math and science), and my profession allows me to blend technology and humanism. I’m in technology because I find its workings and possibilities interesting; I’m in usability because it allows me to find deeper meaning in my work, knowing that I’m helping make people’s lives easier. That’s classic NF stuff. But I think my writing reflects pretty well who I am, especially considering that I’m blogging about my professional field and not my personal life. I’ll give Typealyzer the benefit of the doubt and award it an A-minus.

I’ve run Typealyzer by a dozen or so INFP friends, and most of them say that it gets them wrong, sometimes very wrong. One thinks it’s because they aren’t blogging about their deepest selves. She may be right — and it makes sense to me — but without knowing how Typealyzer works I really can’t say.

King asks people to let her know whether their blogs match their personality types. Here’s what I wrote:

Mine was very close. From my blog (http://www.luminanze.com/blog), Typealyzer classified me as an INTP — and I am an INFP… with strong T skills. I’d say that my writing reflects pretty well who I am, and I give Typealyzer an A-minus.

I would find it very interesting to see a study of a large number of blogs in many fields, with the aim of teasing out the factors that can contribute to the discrepancies found.

To Jim Profit’s comment (“of course it fits, because no matter what sign you are the predictions are so vague they always fit”) — Jim, that’s just not so. Type is not about predicting the future, it’s about describing how a person relates to the world. I find commonality with more than one description — INFP and INTP, in particular (and to a lesser extent INFJ) — but in no way, shape, or form do I find that the ESTJ profile (just to take one example) describes anything remotely close to who I am.

So, about that study…

Here’s what I’d like to see:

Large number of blogs. Depends on the number of hypotheses tested (see below), but I would expect it to need at least ten blogs in each field. Maybe twenty.

Representative sample. This is going to be very hard to achieve. It can’t be done by broadcasting a request asking for volunteers. Self selection is one of the worst ways to recruit a sample that truly represents the population of interest, especially when you’re addressing issues of personality. (Some personalities are more likely to volunteer, eh?) So recruitment would need to be done by invitation. Perhaps the blogs could be submitted to Typealyzer and then the authors would be invited to take the Myers-Briggs (or related online questionnaire, although AFAIK none of those have been validated — I mean, we’re talking about a valid study here). However, the researchers would have to refrain from looking at the Typealyzer results before issuing the invitations.

Many fields. Not being an expert on the blogosphere, I can’t say off the top of my head how many fields the study should include, or which ones. I can only say right now that its designers would need to analyze the blogosphere and construct a categorization scheme that itself is representative of the blogosphere.

Teasing out the factors. This is all about hypothesis development. Can’t really be specified in a short blog post, without substantial analysis a priori. But I would expect it to include questions that would get at the four Ps: whether the blog was personal, professional, political, or playful.

Data collection. Just an online survey, I think. Except for the type inventory.

Whew, that’s a tall order!

Anyone up for collaborating?