Trying to get a little more quantifiable with my social research.

[A. Ashkuff]

Trying to get a little more quantifiable with my social research.

So I broke out some statistical mathematics, and applied it to the New Age, Vedic, Christian, and Pagan spirituality I observed in a farmers market. I'd love to hear any commentary on my methods, as I'm still new to stats.

 

http://ashkuff.com/akaBlog/?p=5077

 

For those of you into photography, and don't care about math, plenty of neat artifacts abounded! Check it out and get back to me.

 

Seriously, fellow social scientists! Let's stay social.

I cannot improve without good input.

 

--- Ashkuff | http://www.ashkuff.com | How to venture out of “armchair” scholarship, and into action? One anthropologist tackles occultism, violence, and more! He gets spooked and roughed up a lot.

[John McCreery]

Off the top of my head,you are dealing with small, non-random samples in which you have discovered a consistent but surprising result. We must note, moreover, that you have only data for one answer to one question for each of the individuals in the sample. In terms of statistical inference, there is not a lot you can do with this. So the question becomes one of relating the finding to the context in which the research was done. Commentors on your blog have already mentioned some possibilities. Is there any obvious reason why people who visit these markets should see spirituality as very important in their lives? If these were church bazaars,the answer would be obvious. What you found might also be the result of a snowball effect. Suppose that the markets were started by people with strongly spiritual leanings. If they attracted customers with similar leanings but put off prospects who find overt spirituality distasteful, the predictable result would shortly be precisely what you find. One might also consider "local" culture and wonder how big the locality is. The neighborhood? The city? The state? The region? Confined to a limited age group or educational stratum.

All this is why one question per subject surveys aren't worth much except as tests of very precisely defined hypotheses, e.g., "I think she loves me. I'll ask her to marry me." One question, one answer, one subject. The result is definitive. If things are at all murky, we need more questions, bigger random samples or both.