Interpreting thousands of multiple-choice demographic surveys

Hi everyone-

I'm trying to analyze a huge batch of interesting demographic data that
I've collected from a network of digital devices in locations across
the USA. Basically, I had consumers answer a few multiple choice
questions like "what is the age of people in this location? a. 21 or
younger, b. 21-29, c. 29-39, d. 39-49, e. 49+ "

The responses are grouped by location. So, for each location, I have
discrete counts of answers. Some people probably just enter random
meaningless answers, but with a lot of responses, you start to see
clusters around certain answers, and that is probably
a good indicator of the actual demographics of that location.

For example, one location might have 10 "a" answers, 40 "b" answers, 30
"c" answers, 15 "d" answers, and 5 "e" answers. By eyeballing these
figures, it's clear that the mean answer is around "b" but both "b" and
"c" are strong and probably define the demographics of
the location (21 - 39 years old).

Is there any straightforward mathmatical way to classify each
collection of answers and give a measure of confidence? i.e., for each
"row" of data, I'd like to be able to identify the 1 or 2 strongest
answers and give some indication of the variance around those answers.

Basically, I've got a series of frequency distributions of discrete
data, and I guess I'm
looking for some kind of a mean and standard deviation for each set.

Any ideas?

Thanks,
Dave

.