In the data processing world (especially as it applies to marketing) there is a very popular term known as “Segmentation”. For some reason, the census community does not seem to have embraced the term, though they use the concept every time they look at particular geographies, genders, races, age groups, … etc.
The idea behind segmentation is that I want to study a particular segment of the population, such as 25-35 year olds that live in single-family homes in Florida and make between $41,000 and $83,000 per year. When the telephone company asked me to create a tool that could answer any question they could think of (regarding customer usage of products and services), segmentation was clearly the only approach that would serve their purposes.
We see a simple form of segmentation (though it isn’t called segmentation) in the Census Bureau’s “Data Ferrett” tool. A word about Data Ferrett first: The Census Bureau is not required to provide you with tools to analyze data, but to help people out, they produced Data Ferrett. They aren’t trying to provide a tool to answer all your questions with Data Ferrett. – in fact they tell you up front that this tool is for people who don’t have access to SAS, clearly implying that SAS would be a better choice if you do have access to it. I’ve never used SAS myself, but SAS is what the telephone company used 20 years ago to answer their questions before I wrote MAST. (SAS quickly fell into the background afterwards.)
The theory behind “answering nearly any question you can produce” via segmentation has been well supported over the years. It’s used not just in MAST, but is the basis for OLAP tools and data mining. Try for a moment to think of information that can be derived from existing data, but can’t be answered via segmentation … if you can think of one quickly, you may not fully understand the capabilities of segmentation – or MAST’s abilities to create new dimensions and volumes (from existing information) used in segmentation. A few months after MAST went live in 1996, I received a call from a very happy telecom manager who emphatically stated: “We haven’t found anything that we can’t do with it yet!”